KO KNOWLEDGE ORGANIZATION Contents - Ergon-Verlag

Knowl. Org. 47(2020)No.1

KO KNOWLEDGE ORGANIZATION Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444

International Journal devoted to Concept Theory, Classification, Indexing and Knowledge Representation

Contents

Peer Review in 2019 ....................................................................... 3 Editorial Richard P. Smiraglia. Referencing as Evidentiary: An Editorial ...................................... 4 Articles Hyerim Cho, Thomas Disher, Wan-Chen Lee, Stephen A. Keating, and Jin Ha Lee. Facet Analysis of Anime Genres: The Challenges of Defining Genre Information for Popular Cultural Objects .............................................................................. 13 Shiv Shakti Ghosh, Subhashis Das, and Sunil Kumar Chatterjee. Human-centric Faceted Approach for Ontology Construction .................................................................. 31 Andrew MacFarlane, Sondess Missaoui, and Sylwia Frankowska-Takhari. On Machine Learning and Knowledge Organization in Multimedia Information Retrieval ........................................... 45 Bartłomiej Włodarczyk . KABA Subject Headings and the National Library of Poland Descriptors in Light of Wojciech Wrzosek’s Theory of Historiographical Metaphors and Different Historiographical Traditions ......................................................... 56

Reviews of Concepts in Knowledge Organization Deborah Lee. Hornbostel-Sachs Classification of Musical Instruments ..................................................................................... 72 Letters to the Editor Claudio Gnoli and Edoardo Manelli. Popularity of entries in ISKO Encyclopedia of Knowledge Organization...................................................................... 92 Guangyue Wei. Databases should Keep Pace with the Needs of scientific Exploration: “Nationality” should be added to scientific Research Databases....................................... 93 Books Recently Published ........................................................ 95 Index to Volume 46 ..................................................................... 96


KNOWLEDGE ORGANIZATION KO Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444


KNOWLEDGE ORGANIZATION This journal is the organ of the INTERNATIONAL SOCIETY FOR KNOWLEDGE ORGANIZATION (General Secretariat: Amos DA-VID, Université de Lorraine, 3 place Godefroy de Bouillon, BP 3397, 54015 Nancy Cedex, France. E-mail: [email protected]. Editors Richard P. SMIRAGLIA (Editor-in-Chief), Institute for Knowledge Or-ganization and Structure, Lake Oswego OR USA 97035. E-mail: [email protected] Joshua HENRY, Institute for Knowledge Organization and Structure, Lake Oswego OR USA 97035. Peter TURNER, Institute for Knowledge Organization and Culture, Lake Oswego OR USA 97035. J. Bradford YOUNG (Bibliographic Consultant), Institute for Knowledge Organization and Structure, Lake Oswego OR USA 97035. Editor Emerita Hope A. OLSON, School of Information Studies, University of Wiscon-sin-Milwaukee, Milwaukee, Northwest Quad Building B, 2025 E New-port St., Milwaukee, WI 53211 USA. E-mail: [email protected] Series Editors Birger HJØRLAND (Reviews of Concepts in Knowledge Organization), Department of Information Studies, University of Copenhagen. E-Mail: [email protected] María J. LÓPEZ-HUERTAS (Research Trajectories in Knowledge Organization), Universidad de Granada, Facultad de Biblioteconomía y Documentación, Campus Universitario de Cartuja, Biblioteca del Colegio Máximo de Cartuja, 18071 Granada, Spain. E-mail: [email protected] Editorial Board Thomas DOUSA, The University of Chicago Libraries, 1100 E 57th St, Chicago, IL 60637 USA. E-mail: [email protected] Melodie J. FOX, Institute for Knowledge Organization and Structure, Lake Oswego OR USA 97035. E-mail: [email protected]. Jonathan FURNER, Graduate School of Education & Information Stud-ies, University of California, Los Angeles, 300 Young Dr. N, Mailbox 951520, Los Angeles, CA 90095-1520, USA. E-mail: [email protected] Claudio GNOLI, University of Pavia, Science and Technology Library, via Ferrata 1, I-27100 Pavia, Italy. E-mail: [email protected] Ann M. GRAF, School of Library and Information Science, Simmons University, 300 The Fenway, Boston, MA 02115 USA. E-mail: [email protected] Jane GREENBERG, College of Computing & Informatics, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104 USA, E-mail: [email protected]

José Augusto Chaves GUIMARÃES, Departamento de Ciência da Informacão, Universidade Estadual Paulista–UNESP, Av. Hygino Muzzi Filho 737, 17525-900 Marília SP Brazil. E-mail: [email protected] Michael KLEINEBERG, Humboldt-Universität zu Berlin, Unter den Linden 6, D-10099 Berlin. E-mail: [email protected] Kathryn LA BARRE, School of Information Sciences, University of Illi-nois at Urbana-Champaign, 501 E. Daniel Street, MC-493, Champaign, IL 61820-6211 USA. E-mail: [email protected] Devika P. MADALLI, Documentation Research and Training Centre (DRTC) Indian Statistical Institute (ISI), Bangalore 560 059, India. E-mail: [email protected] Daniel MARTÍNEZ-ÁVILA, Departamento de Ciência da Informação, Universidade Estadual Paulista–UNESP, Av. Hygino Muzzi Filho 737, 17525-900 Marília SP Brazil. E-mail: [email protected] Widad MUSTAFA el HADI, Université Charles de Gaulle Lille 3, URF IDIST, Domaine du Pont de Bois, Villeneuve d’Ascq 59653, France. E-mail: [email protected] H. Peter OHLY, Prinzenstr. 179, D-53175 Bonn, Germany. E-mail: [email protected] M. Cristina PATTUELLI, School of Information, Pratt Institute, 144 W. 14th Street, New York, New York 10011, USA. E-mail: [email protected] K. S. RAGHAVAN, Member-Secretary, Sarada Ranganathan Endowment for Library Science, PES Institute of Technology, 100 Feet Ring Road, BSK 3rd Stage, Bangalore 560085, India. E-mail: [email protected]. Heather Moulaison SANDY, The iSchool at the University of Missouri, 303 Townsend Hall, Columbia, MO 65211, USA. E-mail: [email protected] M. P. SATIJA, Guru Nanak Dev University, School of Library and Infor-mation Science, Amritsar-143 005, India. E-mail: [email protected] Aida SLAVIC, UDC Consortium, PO Box 90407, 2509 LK The Hague, The Netherlands. E-mail: [email protected] Renato R. SOUZA, Applied Mathematics School, Getulio Vargas Foundation, Praia de Botafogo, 190, 3o andar, Rio de Janeiro, RJ, 22250-900, Brazil. E-mail: [email protected] Rick SZOSTAK, University of Alberta, Department of Economics, 4 Edmonton, Alberta, Canada, T6G 2H4. E-mail: [email protected] Joseph T. TENNIS, The Information School of the University of Wash-ington, Box 352840, Mary Gates Hall Ste 370, Seattle WA 98195-2840 USA. E-mail: [email protected] Yejun WU, School of Library and Information Science, Louisiana State University, 267 Coates Hall, Baton Rouge, LA 70803 USA. E-mail: [email protected] Maja ŽUMER, Faculty of Arts, University of Ljubljana, Askerceva 2, Ljubljana 1000 Slovenia. E-mail: [email protected]


Peer Review in 2019

3

Peer Review in 2019

DOI:10.5771/0943-7444-2020-1-3

Knowledge Organization, like all scientific journals, relies on extensive peer review of manuscript submissions to ensure the originality and quality of research published in our pages. On behalf of the KO Editorial Board we would like to express our unqualified gratitude to the scholars listed below who provided peer review during 2019. Alkim Akdag Salah Danielle Allard Frederik Åström Greg Bak Betty Beunk Thiago Bragato Barros Vanda Broughton John Budd Andrew Buxton Shu-Jiun Chen Hans Dam Christensen Paul Cleverley Gerard Coen Sangeeta Deokattey Stella G. Dextre Clarke Thomas Dousa Melodie Fox Martin Frické Alon Friedman Jonathan Furner Isidoro Gil-Leiva Melissa Gill Claudio Gnoli Koraljka Golub Ann Graf Jane Greenberg Richard Griscom José Augusto Chaves Guimarães Renata Gutierres Castanha Isto Huvila Gregory H. Leazer Fidelia Ibekwe-SanJuan David Jank Nattapong Kaewboonma Ulrika Kjellman Michael Kleineberg Kathryn La Barre María J. López-Huertas Robert Losee Kun Lu

Niels D. Lund Devika Madalli Fulvio Mazzocchi Carlos Marcondes Daniel Martínez-Ávila Shawne Miksa Elizabeth Milonas Robert Montoya Hyoungjoo Park Daniel Parrochia Christina Pattuelli Emma Quinlan Pauline Rafferty Richard A. Richards Maja Rudloff Laura Ridenour Riccardo Ridi Steven Robertson Renato Rocha Souza Mohinder P. Satija Andrea Scharnhorst Arthur Smith Rosale Souza Louise Spiteri Rick Szostak Joseph T. Tennis Martin Thellefsen Natalia Tognoli Garry Trompf Douglas Tudhope Hannah Turner Herbert van de Sompel Kim Veltman Judi Vernau Howard D. White Yejun Wu Marcia Zeng Alesia Ann Zuccala Maja Žumer


R. P. Smiraglia. “Referencing as Evidentiary: An Editorial

4

Referencing as Evidentiary: An Editorial

Richard P. Smiraglia

Institute for Knowledge Organization and Structure, Lake Oswego OR 97035 USA, <E-mail: [email protected]>

Richard P. Smiraglia holds a PhD in information from the University of Chicago. He is Senior Fellow and Ex-ecutive Director of the Institute for Knowledge Organization and Structure, Inc. and is Editor-in-Chief of this journal. He also is Professor Emeritus of the iSchool at the University of Wisconsin-Milwaukee. He was 2017-2018 KNAW Visiting Professor at DANS (Data Archiving and Networked Services division of the Royal Netherlands Academy of the Arts and Sciences), The Hague, The Netherlands, where he remains visiting fel-low and was the 2018 recipient of the 2018 Frederick G. Kilgour Award for Research in Library and Infor-mation Technology.

Smiraglia, Richard P. 2020. “Referencing as Evidentiary: An Editorial.” Knowledge Organization 47(1): 4-12. 9 ref-erences. DOI:10.5771/0943-7444-2020-1-4.

Abstract: The referencing habits of scholars, having abandoned physical bibliography for harvesting of digital resources, are in crisis, endangering the bibliographical infrastructure supporting the domain of knowledge organization. Research must be carefully managed and its circumstances controlled. Bibliographical replicability is one important part of the social role of scholarship. References in Knowledge Organization volume 45 (2018) were compiled and analyzed to help visualize the state of referencing in the KO domain. The dependence of science on the ability to replicate is even more critical in a global distributed digital environment. There is great richness in KO that make it even more critical that our scholarly community tend to the relationship between bibliographical verity and the very replicability that is allowing the field to grow theoretically over time.

Keywords: ISKO international conference, knowledge organization, citations, references

1.0 Referencing versus harvesting That there is a crisis—looming or already upon us—in the referencing habits of scholars is obvious from a quick glance at any reference list in any article in any scientific journal. The once noble chore of jotting down the ap-propriate bibliographical characteristics of any work con-sulted in scholarship—a practice that guaranteed a sort of replicability—has given way to the harvesting of cita-tions from online resources. The notion that a scholar must cite a source that other scholars also can consult has given way to various forms of ritual citation. The danger in ritual citation goes far beyond the obvious inaccuracy of references or the time wasted by scholars chasing dead ends looking to read a particularly interestingly cited piece of prior work. The danger lies in the collapse of the bibliographical infrastructure that supports all schol-arship.

My mentor at the University of Chicago, Abraham Bookstein, famously opened his research methods course with the statement that all research is consciously pre-meditated inquiry (1982):

If we are seriously concerned with obtaining evi-dence that can substantiate our beliefs about some subject area, then we must develop approaches that yield information that is as valid and persuasive as possible. To the extent that the evidence we pro-duce is compatible with our beliefs, and to the ex-tent that alternative explanations for the evidence seem implausible, to that extent our research effort is productive. By subjecting our beliefs to a range of careful tests, we are able to expose ideas that are faulty and to gain confidence in those that are valid. In this way our comprehension of a substantive ar-ea is built up.

Bookstein’s two points intermingle. The first is that re-search is self-conscious inquiry, which means the scholar manages every aspect of the research question and every aspect of evidence that can be applied to the answer. The second and related point is that research is premeditated, planned to control the circumstances by which evidence is gathered, analyzed and synthesized.1



5

2.0 The social role of bibliographical replicability Another mentor, D. Kathryn Weintraub, was my cataloging professor at Indiana University in 1973-74. We had a pretty good rapport, and one day when I visited her office to dis-cuss a paper I was writing about Brown’s classification (e.g., see Beghtol 2004) she took the opportunity to show me how to work with research resources. I had brought a stack of things from the library with me. Dr. Weintraub put some paper in her typewriter, opened my first source, and typed out a perfect Chicago/Turabian reference based on what she could see by looking at the resource. Then she skipped a couple of spaces and typed a page number and started typing out passages I had highlighted—a practice I still follow today, albeit using a computer. It means, of course, that there is perfect replicability so long as the quo-tation is accurate and the resource citation is precise. There was no internet, there was only the actual evidence itself in printed form in peer-reviewed journals. It was a perfect system, but vulnerable, as we have learned subsequently, to inefficiencies of scale. Still, the lesson she taught me that day was that as a scholar I was part of a community that shared responsibility for the growth of knowledge. There was a clear social dimension.

That there is a social dimension to consciously premedi-tated inquiry almost goes without saying. Obviously, evi-dence must be acceptable across social boundaries, which means that phenomenological philosophical prerogatives apply to the interpretation of data, as well as to any con-clusions from any particular act of consciously premeditat-ed inquiry and stretch on into the extension of theoretical conclusions demonstrated by repeated hypothesis testing. Nagel (1979, 495 emphasis original) reminds us that “val-ue-free social science is impossible, because value com-mitments enter into the very assessment of evidence.” One hopes we all learned these tenets as beginning researchers and we all bring conscious premeditation to the statement of hypotheses derived from prior research, to the design of research methods, especially experiments, and to the construction of analytical tools and procedures. But what of the bibliographical parameters of research?

An important point to stress repeatedly in today’s wired scholarly world is the importance of the notion that evidence must be replicable, and that bibliographic infrastructure is essential evidence. Incorrect references obfuscate replication by preventing any future scholar from following the tracks of a predecessor. References must be discoverable. Bibliographic replicability demands precision and accuracy—a form of scholarship once no-bly known as “bibliography.” As Krummel (1984, 9) re-minded us “any text that is significant and substantial enough to be published ought to be known about so that it can be consulted,” and “a text that is not discoverable,

like the one that is not available, for all practical purposes does not exist.”

As it happens, even in the twenty-first century biblio-graphical accuracy is more than a matter of cutting and pasting from the “cite as” link on a website. Krummel describes the act of bibliographical description as the capturing of Platonic realities (25):

Any item appropriate to a list will by its nature have two Platonic realities. It occupies space, and will continue to do so through forseeable time, thus as-suming a physical form. It also consists of con-tent—that is, a message, verbal, intellectual, artistic, or spiritual (however such terms may be defined), sent by an original creator, modified by intermedi-ary producers, and eventually perceived by an audi-ence of readers … The title, for instance, describes the content and at the same time names the physi-cal object; the imprint, by identifying the producer of the physical object, also tells where both the content and the physical object can, or at one time could, be obtained.

Thus, in the accurate reproduction of bibliographic data, scholarship has the opportunity to demonstrate the evi-dentiary role of a specific text in space and time, on the one hand, and to provide a direct pathway to its retrieval, and thus to its usefulness in replication, on the other.

In 2017 I wrote in this journal about the need for greater replication in knowledge organization research. It is a tenet of science that the organization and classifica-tion of discoveries is at the basis of any theory. Theory cannot grow from single isolated observations, but rather is always the result of the synthesis of accumulated ob-servations undertaken across scholarly domains over time. I concluded (317): “science relies not on spiritual warrant, not even on common human sense, but rather on replicated and replicable and therefore empirically ver-ifiable controlled observation, the results of which are classified.” To get to that halcyon place in the evolution of the domain requires replicable evidence, and the de-mand for replicable evidence requires bibliographical precision. References are evidence, and evidence must be both precise and replicable. 2.1 What is a scholar to do? Our editorial policy includes verification of every refer-ence in every manuscript published in our journal. The scientific position we take with that policy is that we must verify the veracity of the cited evidence as it enters through our journal into the published science of knowledge organization (KO). Our journal is the bench-



6

mark for the domain of KO and we take our gatekeeping function seriously. We verify every citation—not just the formatting, but the facts (the Platonic realities as Krummel might have said). We are not alone in this—I recall having a very red metaphorical face in my earliest career as a scholar when a journal copy editor sent me two of my own citations that could not be verified and asked me to either correct them or provide evidence to support their veracity. Of course, I had made an error (in the dates of publication as I recall). With chagrin but also relief I made the corrections. Not all that long ago it happened again, in a joint submission for which I had left the references to a collaborator and the journal editors easily identified several errant references. Thus, it is a regular part of the responsibility of scientific publishing to maintain a high level of accuracy in references to work cited. The references are the direct routes to the authors’ evidence, after all, as I hope I demonstrated earlier.

The majority of references by the majority of authors are accurate or require only slight emendation. That is the good news. Problematic references fall easily into two cat-egories: 1) those for which it becomes clear the author has not actually consulted the source and has instead copied a reference from some other paper; and 2) conference pa-pers in online proceedings. Obviously, these categories dif-fer substantively. As for the first group, it is our editorial policy to inquire of authors, when necessary, exactly which text of the work cited has been consulted, and then we help to build or correct the reference from that document.

Conference papers are another matter altogether. The model for a reference for a conference paper derives from the once common practice of publishing print volumes of proceedings for all conference participants. Such volumes constitute anthologies, and the papers in them can be cit-ed thus: Schallier, Wouter. 2004. “On the Razor’s Edge: Be-

tween Local and Overall Needs in Knowledge Organi-zation.” In Knowledge Organization and the Global Infor-mation Society: Proceedings of the Eighth International ISKO Conference 13-16 July 2004 London, UK, ed. Ia C. McIl-waine. Advances in Knowledge Organization 9. Würz-burg: Ergon Verlag, 269-74.

Note the elegant detail: author, date, article title, complete ti-tle proper of proceedings volume in which it occurs, state-ments of responsibility for the proceedings, series statement, place, publisher and exact page numbers. All of the principal personalities identified and their relationships for the physi-cal artifact clarified all in one succinct statement.

Alas, scholarly societies increasingly are abandoning the expensive production of printed volumes in lieu of online proceedings, which are more than adequate for the task of

providing recent content to participants in digital form. But the “published” proceedings then frequently lack even title pages, the title of the conference might vary from place to place in the digital resource, and details of publica-tion (i.e., such niceties as place of publication and publish-er name) often are missing altogether. The venerable Asso-ciation for Computing Machinery (ACM) has a massive digital library containing many sets of proceedings, and they thus provide (responsibly, I might add) detailed refer-ence data. Here is an example: Zubiaga, Arkaitz, Christian Körner and Markus

Strohmaier. 2011. “Tags vs Shelves: from Social Tag-ging to Social Classification.” In HT'11:Proceedings of the 22nd ACM Conference on Hypertext and Hypermedia; June 6-9, 2011, Eindhoven, the Netherlands. New York, NY: Association for Computing Machinery, 93-102. doi:10.1145/1995966.1995981

But there also is the problem that many papers presented at conferences are not actually published in online pro-ceedings, and thus should be identified not as “In Proceed-ings ….” but rather as “Unpublished paper presented at ….” As Krummel reminded us, to generate a reference for “Proceedings ….” is to create a sort of bibliographic ghost—a reference to a physical or digital resource that actually never existed and therefore cannot be consulted. 3.0 One year of KO references: some data, some

comments A few observations about the role of bibliographical ref-erences in our journal can be garnered from a brief over-view of the references in a recent volume. All references in the journal in volume 45 from 2018 were compiled for this purpose. Let us state for the record that our journal uses the author-date referencing system, so each “work” only occurs in a reference list once per article, no matter how many times it might be “referenced” from within that article’s text. As this is not a domain analysis, but ra-ther simply some descriptive research to help illuminate the state of referencing in our journal, only a few param-eters are reported. In addition to the commentary pro-vided here, good readers are invited—welcomed even—to consider the references in the examples below to be standard and reusable for the core literature cited.

There were 2,525 references in 44 articles, of which 32 were regularly contributed articles and 12 were review ar-ticles contributed by the online ISKO Encyclopedia of Knowledge Organization (https://www.isko.org/cyclo/). The mean number of references per article was 60.56; the median and mode were both 43. However, the articles from IEKO are review articles, with much more detailed



7

reference lists. The mean number of references per re-view article was 127, the median was 81 and the mode was 120.5. The range of dates of publication stretched from 1635 to 2018. The mean age of citation was 19 years; the median was 11 years; and the mode was 2 years. Thus, most of the references were to very recent re-search, but a large component also referenced core theo-retical material. Another way of looking at this is to con-sider that most of the citations are not to classical texts for which standard bibliographic details are well-known. Rather, most of the citations were to recent publications in online journals and conference proceedings—perilous territory for bibliographic veracity.

Further evidence of this trend arises from the fact that there are 1,935 singleton references—works by authors who are not cited again in the corpus. We recognize, of course, the outlines of a standard Bradford-like power dis-tribution—most contributions are singletons, a very few represent works from the members of a more oft-cited re-search community. Three quarters of the references (76.6%) in this single volume of our journal, then, were to

works by authors that were cited only once. A small body of works by authors cited more than once (23.3%) might point to the presence of a theoretical core. In fact, there were 590 references to authors’ whose names appear more than once. 179 authors appeared twice, 42 appeared three times, and 28 appeared 4 times. Table 1 shows those 75 au-thors whose names appear 5 times or more.

It is difficult to know where to draw a line across seg-ments of this frequency distribution. Obviously, those named with the highest frequencies are the most influen-tial, at least in this particular volume year of the journal. We might also hazard a guess that the group of authors at the top of the distribution, at least in 2018, were defining the domain of KO. But the whole list itself is also very in-formative, including names of classical (and ritually-cited) authors from the history of the related field of infor-mation science.

Greater insight comes from analysis of the works rep-resented. Of the 75 oft-cited authors in Table 1, only a few represent specific works that are cited 3 or more times. These are shown in Table 2.

Cited Author Frequency

Hjørland, Birger 83

Smiraglia, Richard P 34

Gnoli, Claudio 26

Buckland, Michael K 21

Mai, Jens Erik 19

Dahlberg, Ingetraut 18

Olson, Hope A 17

Shera, Jesse H 14

Frohmann, Bernd; Mazzocchi, Fulvio 13

Beghtol, Clare; Guimarães, José Augusto Chaves; Otlet, Paul; Ranganathan, S; Vickery, B 12

Lancaster, Frederick Wilfrid 11

Furner, Jonathan; Lund, Niels Windfeld; Soergel, Dagobert; Zeng, Marcia Lei 10

Floridi, Luciano; Rafferty, Pauline 9

Andersen, Jack; Berners-Lee, Tim; Briet, Suzanne; Broughton, Vanda; Chan, Lois Mai; Fox, Melodie J; Frické, Martin 2009; Fugmann, Robert; IPCC (Intergovernmental Panel on Climate Change) 8

Bates, Marcia J; Berman, Sanford; Bliss, Henry Evelyn; Eco, Umberto; Foskett, Douglas John; Golub, Koraljka; Miksa, Francis; Oh, Dong-Geun; Rayward, W; Svenonius, Elaine; Wilson, Patrick

7

Anderson, James D; Barité, Mario; Bawden, David; Chen, Chaomei; Cleverdon, Ciryl W; Ibekwe-SanJuan, Fidelia; In-ternational Organization for Standardization (ISO); Kuhn, Thomas S; Library of Congress; Peters, Isabella; Satija, Mohinder Partap

6

Borgman, Christine L; Ellis, David; Farradane, Jason; Fouillée, Alfred; Fuller, Steve W; García Gutiérrez, Antonio 2002; Grolier, Éric de; Gross, Tina; Kipp, Margaret E; Leonelli, Sabina; Munk, Timme Bisgaard; Nelson, Theodor Holm; Ørom, Anders; Rowley, Jennifer; Slavic, Aida; Small, Henry G; Swanson, Don R; Van Rijsbergen, C; Weinberg, Bella Hass; White, Howard D; Zhao, Dangzhi; Žumer, Maja

5

Table 1. Authors cited 5 or more times.



8

Cited work Frequency

Briet, Suzanne. 1951. Qu’est-ce que la documentation? Collection de documentologie 1. Paris: Éditions documentaires, indus-trielles et techniques. – [includes 2 translations]

7

Hjørland, Birger 2017a. “Classification.” Knowledge Organization 44: 97-128. 6

Hjørland, Birger. 2008. “What is Knowledge Organization (KO)?” Knowledge Organization 35: 86-100. 6

Beghtol, Clare. 2002. “A Proposed Ethical Warrant for Global Knowledge Representation and Organization Sys-tems.” Journal of Documentation 58: 507-32.

6

Bliss, Henry E. 1929. The Organization of Knowledge and the System of the Sciences. New York, NY: Henry Holt and Company. 6

Kuhn, Thomas S. 1962. The Structure of Scientific Revolutions. Chicago, IL: University of Chicago Press. 6

Lancaster, F. W. 2003. Indexing and Abstracting in Theory and Practice. London: Facet Publishing. – [includes 3 eds.] 5

Shera, Jesse H. 1951. “Classification as the Basis of Bibliographic Organization.” In Bibliographic Organization: Papers Presented Before the Fifteenth Annual Conference of the Graduate Library School July 24-29, 1950, ed. Jesse H. Shera and Margaret E. Egan. The University of Chicago Studies in Library Science. Chicago: University of Chicago Press, 72-93.

5

Otlet, Paul. 1934. Traité de documentation: Le livre sur le livre, thee ́orie et pratique. Bruxelles: Editiones Mundaneum. 5

Ranganathan, S. R. 1967. Prolegomena to Library Classification, 3rd ed. Ranganathan Series in Library Science 20. Bombay: Asia Publishing House.

5

Vickery, Brian C. 1956. “Notational Symbols in Classification. Part 2: Notation as an Ordering Device.” Journal of Docu-mentation 12: 73-87. 5

Zeng, Marcia Lei. 2008. “Knowledge Organization Systems (KOS).” Knowledge Organization 35, no 2: 160–82. 5

Olson, Hope A. 2002. The Power to Name: Locating the Limits of Subject Representation in Libraries. Dordrecht, The Nether-lands: Kluwer Academic Publishers.

4

Mazzocchi, Fulvio. 2018. “Knowledge Organization System (KOS): An Introductory Critical Account.” Knowledge Organ-ization 45: 54-78.

4

Frické, Martin 2009. “The Knowledge Pyramid: A Critique of the DIKW Hierarchy.” Journal of Information Science 35: 131-42.

4

Wilson, Patrick. 1983. Second-hand Knowledge: An Inquiry into Cognitive Authority. Contributions in Librarianship and Infor-mation Science 44. Westport, CT: Greenwood.

4

Ibekwe-SanJuan, Fidelia and Geoffrey C. Bowker. 2017. "Implications of Big Data for Knowledge Organization.” Knowledge Organization 44: 187-98.

4

Smiraglia, Richard P. 2015. Domain Analysis for Knowledge Organization: Tools for Ontology Extraction. Chandos Information Professional Series. Waltham, MA: Chandos Publishing.

4

Mai, Jens-Erik. 2011. “Folksonomies and the New Order: Authority in the Digital Disorder.” Knowledge Organization 38: 114-22.

4

Mai, Jens-Erik. 2013. “Ethics, Values and Morality in Contemporary Library Classifications.” Knowledge Organization 40: 242-53.

4

Mai, Jens-Erik. 2011. “The Modernity of Classification.” Journal of Documentation 67: 710-30. 4

Hjørland, Birger and Albrechtsen, Hannah. 1995. “Toward a New Horizon in Information Science: Domain Analysis.” Journal of the American Society for Information Science 46: 400-25. 3

Hjørland, Birger. 2012. “Is Classification Necessary after Google?” Journal of Documentation 68, 3: 299-317. 3

Hjørland, Birger. 2015. “Theories are Knowledge Organizing Systems (KOS).” Knowledge Organization 42: 113-28. 3

Hjørland, Birger. 2015a. “Classical Databases and Knowledge Organization: A Case for Boolean Human Decision-making During Searches.” Journal of the Association for Information Science and Technology 66: 1559-75. 3

Hjørland, Birger. 2016. “Informetrics Needs a Foundation in the Theory of Science.” In Theories of Informetrics and Scholar-ly Communication: A Festschrift in Honor of Blaise Cronin, ed. Cassidy R. Sugimoto. Berlin: Walter de Gruyter, 20-46

3

Dahlberg, Ingetraut. 1995. “Current Trends in Knowledge Organization.” In Organización del conocimiento en siste-mas de información y documentación : actas del II Encuentro de ISKO-España, Getafe, 16 y 17 de noviembre de 1995, ed. Francisco J. Garcia Marco. Zaragoza: Universidad de Zaragoza, 7-26.

3

Frohmann, Bernd. 1990. “Rules of Indexing: A Critique of Mentalism in Information Retrieval Theory.” Journal of Docu-mentation 46, no. 2: 81-101.

3

Table 2. Works cited 3 or more times. (continued on next page)



9

Here we have fifty-one “core” articles. (And, we might note also, here we have fifty-one properly formatted and verified references for those core works.) That is, in the volume year 2018, these fifty-one papers are considered to be influential, and their influence is demonstrated by the citation frequency. But, on the other hand, dispersion is

more common than clustering. For example, of 83 cita-tions to work by Hjørland, 27 are to 8 specific works cited 3 to 6 times.

Figures 1 and 2 provide visualizations of the works that are

cited the most.

Cited work FrequencyFrohmann, Bernd. 1990. “Rules of Indexing: A Critique of Mentalism in Information Retrieval Theory.” Journal of Docu-mentation 46, no. 2: 81-101.

3

Mazzocchi, Fulvio. 2013. “Images of Thought and their Relation to Classification: The Tree and the Net.” Knowledge Or-ganization 40: 366-74.

3

Guimaräes, José Augusto Chaves. 2017. “Slanted Knowledge Organization as a New Ethical Perspective.” In The Organ-ization of Knowledge: Caught Between Global Structures and Local Meaning, ed. Jack Andersen and Laura Skouvig. Studies in Information 12. Bingley, UK: Emerald, 87-102.

3

Vickery, Brian C. 1953. “The Significance of John Wilkins in the History of Bibliographical Classification.” Libri 2: 326-343.

3

Lund, Niels Windfeld. 2004. “Documentation in a Complementary Perspective.” In Aware and Responsible: Papers of the 2001 Nordic-International Colloquium on Social and Cultural Awareness and Responsibility in Library, Information and Documentation Studies (SCARLID), ed. W. Boyd Rayward. Lanham, MD: Scarecrow Press, 93-102.

3

Rafferty, Pauline and Rob Hidderley. 2007. “Filckr and Democratic Indexing: Dialogic Approaches to Indexing.” Aslib Proceedings 59: 397-410. doi:10.1108/00012530710817591

3

Broughton, Vanda. 2004. Essential Classification. New York: Neal Schuman. 3

IPCC (Intergovernmental Panel on Climate Change). 1996. Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change; Scientific-Technical Analyses; Contribution of Working Group II to the Second Assessment Report of the Intergovernmental Panel on Climate Change, ed. Robert T. Watson, Marufu C. Zinyowera, and Richard H. Moss. Cambridge: Cambridge University Press.Webster, Peter J., Greg Holland, Judith Curry, and H.R. Chang. 2005. “Changes in Tropical Cyclone Number, Duration, and Intensity in a Warming Environment.” Science 309: 1844-6.

3

Berman, Sanford. 1971. Prejudices and Antipathies: A Tract on the LC Subject Heads Concerning People. Metuchen, N.J.: Scare-crow Press

3

Miksa, Francis. 1998. The DDC, the Universe of Knowledge, and the Post-Modern Library. Albany, NY: Forest Press. 3

Rayward, W. Boyd. 1994. “Visions of Xanadu: Paul Otlet (1868-1944) and Hypertext.” Journal of the American Society for In-formation Science 45: 235-50.

3

Svenonius, Elaine 2000. The Intellectual Foundation of Information Organization. Digital Libraries and Electronic Publishing. Cambridge, MA: MIT Press.

3

Wilson, Patrick. 1968. Two Kinds of Power: An Essay on Bibliographical Control. University of California Publications: Librari-anship 5. Berkeley: University of California Press.

3

Bawden, David and Lyn Robinson. 2012. Introduction to Information Science. London: Facet. 3

Peters, Isabella and Katrin Weller. 2008. “Tag Gardening for Folksonomy Enrichment and Maintenance.” Webology 5, no 3: 1-18.

3

Munk, Timme Bisgaard and Kristian Mørk. 2007. “Folksonomy, the Power Law & the Significance of the Least Effort.” Knowledge Organization 34: 16–33.

3

Rowley, Jennifer and John Farrow. 2016. Organizing Knowledge: An Introduction to Managing Access to Information, 4th ed. Lon-don: Routledge.

3

White, Howard D. and Kate W. McCain. 1998. "Visualizing a Discipline: An Author Co-Citation Analysis of Information Science, 1972-1995." Journal of the American Society for Information Science 49: 327-55.

3

Žumer, Maja. 2017. “IFLA Library Reference Model (LRM): Harmonization of the FRBR Family.” ISKO Encyclopedia of Knowledge Organization. Available at http://www.isko.org/cyclo/lrm

3

Smiraglia, Richard P. 2015. “Domain Analysis of Domain Analysis for Knowledge Organization: Observations on an Emergent Methodological Cluster.” Knowledge Organization 42: 602-11.

3

Buckland, Michael K. 1997. “What is a ‘Document’?” Journal of the American Society for Information Science 48: 804-9. 3

Buckland, Michael. 2018. “Document Theory.” Knowledge Organization 45: 425-36. 3

Table 2. Works cited 3 or more times. (continued from previous page)



10

4.0 On bibliographical verity, replicability, and referencing in KO

As scholarship changes so must its technicalities shift and that means the task of embracing the amazing body of digitally-available cited scholarly literature must somehow be adapted to the task of bibliographical verity. The de-pendence of science on the ability to replicate is even more critical in a global distributed digital environment. I

have not even raised the issue of instantiation here (Smi-raglia 2018) but it is relevant always to ask a scholar ex-actly which “version” of that paper were you reading when you made the synthetic leap that caused you to cite it? The question certainly extends from to the matter of editions of printed texts to the now commonplace prac-tice of producing texts on demand. The Platonic realities of space and physicality notwithstanding, it is incumbent on the KO community to corral the practices of refer-

Figure 1. Briet to Kuhn: the most cited works in 2018.



11

Figure 2. Lancaster to Vickery: the second tier of most cited works in 2018.



12

encing that create the theoretical bibliographical infra-structure of the science of KO. It is worth noting that a large bibliography of KO related texts is maintained by ISKO on its website (“KO Literature” https://www. isko.org/lit.html) but that there is little coherence across that database in the format of references.

We see from this brief glimpse of one year of refer-ences that our science is sufficiently complex to yield in a single volume of a journal a long tail of 1,935 singleton references, and that even those works cited with higher-frequency constitute more of a diverse cluster than a crit-ical core. These are signs of great richness in KO that make it even more critical that our scholarly community tend to the relationship between citation practice (or ref-erencing as I have called it here), bibliographical verity and the very replicability that is allowing the field to grow theoretically over time. This theoretical growth, demon-strated very clearly in the growth of the ISKO Encyclopedia of Knowledge Organization is dependent on continued atten-tion to the bibliographical infrastructure of our domain. Note 1. Abraham Bookstein’s commitment to empirical re-

search, and indeed, to “consciously premeditated in-quiry,” was most influential in my own work in the ex-tension of his application of probabilities of selection in sampling. Famously, his exemplary articles “How to Sample Badly” (1974) and “Sampling from Card Files” (1983) are simple and direct presentations of the im-

portance of conscious premeditation on the problems of generalizing from samples of informetric objects to large populations. He carefully guided me in designing the procedure for sampling works from a population of carriers, for example (see Smiraglia 2001, 155-64 “Sampling Works”).

References Beghtol, Clare. 2004. “James Duff Brown's Subject Clas-

sification and Evaluation Methods for Classification Systems.” Library Trends 25 no. 4: 702-18.

Bookstein, Abraham. 1974. “How to Sample Badly.” Li-brary Quarterly 44: 124-32.

Bookstein, Abraham. 1982. Unpublished statistics course notes recorded by R.P. Smiraglia.

Bookstein, Abraham. 1983. “Sampling from Card Files.” Library Quarterly 53: 307-12.

Krummel, D. W. 1984. Bibliographies: Their Aims and Meth-ods. New York: Mansell Publishing Ltd.

Nagel, Ernest. 1979. The Structure of Science: Problems in the Logic of Scientific Explanation. Indianapolis: Hackett.

Smiraglia, Richard P. 2001. The Nature of a Work: Implications for the Organization of Knowledge. Lanham, Md.: Scarecrow Press.

Smiraglia, Richard P. 2017. “Replication and Accumulation in Knowledge Organization—An Editorial.” Knowledge Organization 44: 315-17.

Smiraglia, Richard P. 2019. “Work.” Knowledge Organization 46: 308-19.


H. Cho, Th. Disher, W.-C. Lee, S. A. Keating and J. H. Lee. Facet Analysis of Anime Genres

13

Facet Analysis of Anime Genres: The Challenges of Defining Genre Information

for Popular Cultural Objects† Hyerim Cho*, Thomas Disher**, Wan-Chen Lee***,

Stephen A. Keating**** and Jin Ha Lee*****

*University of Missouri, 303 Townsend Hall, 221K, Columbia, MO 65211, USA, <[email protected]>

***University of Washington, Information School, Mary Gates Hall, Suite 370, Box 352840, Seattle, WA 98195, USA, <[email protected]>,

****1201 Elm Forest Drive, Cedar Park, TX 78613, USA, < [email protected]> **, *****University of Washington, Information School, Mary Gates Hall 330Y-2 Gamer Lab,

1851 NE Grant Ln., Seattle, WA 98105, USA, **<[email protected]>, *****<[email protected]>

Hyerim Cho earned her PhD in information science from the University of Washing-ton. Focusing on users, her research investigates diverse multimedia and pop-cultural information based on information science and media studies to provide enhanced rec-ommendation and retrieval services. Her research interests include information needs and behavior, metadata, qualitative studies and mixed methods, and visual narrative materials such as graphic novels, anime, and video games.

Thomas Disher is a writer and stay-at-home-father. He holds a BA in English creative writing, with a linguistics minor from Seattle Pacific University and a MLIS from the University of Washington. Since earning his MLIS he has assisted Jin Ha Lee and the GAMER group with some of their research at the University of Washington. His re-search interests include metadata, classification systems, video games, anime, folklore, and linguistics. He also writes poetry and fiction.

Wan-Chen Lee is a PhD candidate at the Information School of the University of Washington, where she received her MLIS degree in 2014. She holds a BA in LIS from National Taiwan University. Her research interests include knowledge organization, classification theory, culture and classification, globalization and localization of cata-loging and classification standards, and metadata.

Stephen Keating is a Senior User Researcher with Dell, with a master’s in information science from the University of Washington, and a master’s in media studies from The New School. Research interests include information organization, metadata, and se-mantic analysis, as well as a particular focus on the preservation of digital ephemera.

Jin Ha Lee is Associate Professor at the University of Washington’s Information School and the director of the GAMER (GAME Research) Group. Her research interests include music, anime, game and multimedia infor-mation seeking and retrieval, information organization and access, and knowledge representation. The GAMER Group explores new ideas and approaches for organizing and providing access to video games and other multi-media, understanding user behavior related to these media, and using them for informal learning. She holds an MS (2002) and a PhD (2008) from the University of Illinois at Urbana-Champaign.

Cho, Hyerim, Thomas Disher, Wan-Chen Lee, Stephen A. Keating and Jin Ha Lee. 2020. “Facet Analysis of Anime Genres: The Challenges of Defining Genre Information for Popular Cultural Objects.” Knowledge Organ-ization 47(1): 13-30. 50 references. DOI:10.5771/0943-7444-2020-1-13.

Abstract: Anime, as a growing form of multimedia, needs a better and more thorough organization for its myriad unique terminologies. Existing studies show patrons’ desire to search and get recommendations for anime. However, due to inadequate indexing and often con-fusing or inaccurate usage of terms, searching and acquiring recommendations remain challenging. Our research seeks to close the gap and


H. Cho, Th. Disher, W.-C. Lee, S. A. Keating and J. H. Lee. Facet Analysis of Anime Genres

14

make discovery and recommendations more viable. In this study, we conducted a facet analysis of anime genre terms that are currently used in thirty-six anime-related English-language databases and websites. Using a card sorting method with an inductive approach to the 1,597 terms collected, we identified and defined nine facets and 153 foci terms that describe different genres of anime. Identified terms can be implemented within different organizational systems including library catalogs, recommendation systems, and online databases to improve genre definitions and search experiences.

Received: 28 December 2017; Revised: 16 July 2018; Accepted: 2 August 2018

Keywords: anime, genre terms, facet analysis, foci terms, Japanese

† This article was originally published in Knowledge Organization v. 45, no. 6 (2018) on pages 484-99 DOI:10.5771/0943-7444-2018-6-484. Inexplicably most of the references were omitted at that time, and the appendix was incorrectly type-set. We apologize most profusely for our error and gratefully republish the article in toto with updated author biographical and contact information.—Editor-in-Chief.

1.0 Introduction Interest in anime, a relatively recent multimedia medium in libraries, continues to increase artistically, commercially, and academically (Exner 2012; Fennell et al. 2013; Ressler 2016). The anime industry has been growing for the last five years (AJA 2016), and the total market value of the anime industry in 2015 was 18.1 billion US dollars, a 12% year on year increase from 2014 (Ressler 2016). Anime is commonly defined as Japanese animation, or an-imation produced in Japan, although to Japanese, “anime covers all animation regardless of provenance” (Rich-mond 2009, 2). Davis (2016, 35) defines it as “Japanese limited-animation, which is a form of animation that uti-lizes fewer frames than the classic Disney animation and a variety of other cost-saving techniques,” focusing on eco-nomic and technical aspects. Brenner (2007, 29) more broadly defines it as:

Animated films produced in Japan for a Japanese au-dience. The word itself comes from the word animeshon, a translation of the English word “anima-tion.” This term encompasses all animated titles in-cluding feature films, television shows, and original video animation (OVA) released to the home enter-tainment market.

Brenner’s definition, particularly “animated films pro-duced in Japan for a Japanese audience,” is debatable due to the fact that it is difficult to draw a clear conclusion about where anime is produced, and, especially today, anime is not only for a Japanese audience but an interna-tional one as well.

Denison (2015, 2 emphasis original) argues that “anime is not just a genre any more than it is simply a kind of ani-mation or a product of only Japanese culture.” Denison ex-pands these limited definitions by saying that “anime needs to be understood more broadly as a cultural phenomenon whose meanings are dependent on context.” This expanded concept of anime views it as a “medium,” “multi-

media,” and “part of a range of (trans)national media cul-tures” (Denison, 22).

Although recently it is becoming more difficult to draw a line between what is anime and what is not, in general, anime has several unique characteristics that are distin-guishable from Western animation. Levi (2013) states that sadness, environmentalism, and terror are unique themes found in anime, and Davis (2016) provides an interview with Robert Napton, a graphic novel writer and former employee at Bandai Entertainment, who states that plot and narrative style are anime’s main appeals to its audience, an audience that includes female adults, who were often excluded in Western comics and animation culture.

Despite the growing prevalence of and demand for anime, current organizational systems for anime are lack-ing, especially when describing genre information. A sur-vey conducted by Exner (2012, 34) showed this clearly, as the majority of library patron respondents expressed “the desire for a greater quantity or variety of support” for ac-cessing anime or “more robust genre cataloguing,” alt-hough some respondents also indicated that they were “categorically not interested in anime or manga through the library.” Strengthening the organizational access points for and information about anime titles would help address such user experience requests and facilitate access to exist-ing collections.

As with other narrative and visual narrative materials, an important access point for anime is genre information. Sardar and Van Loon (2015, 49) define genre as “catego-ries of media products; specific types of books, film, tele-vision and radio programs. Genres are identified by their conventions, which the audience recognizes through regu-lar contact. The term originated in literary criticism but was adopted by cinema studies. Television is exclusively genre-led.” In a previous study on investigating anime us-ers’ information needs for recommendations (Cho et al. 2017), genre was one of the most frequently mentioned features used to get recommendations (ranked third among nineteen features identified). Lee, Shim, and Jett (2015) also had similar findings: genre was the second


H. Cho, Th. Disher, W.-C. Lee, S. A. Keating and J. H. Lee. Facet Analysis of Anime Genres 15

most mentioned feature when Korean anime users sought recommendations.

However, despite such common usage, genre in anime remains highly ambiguous. Even some anime fans may not be familiar with some of the genre terms in the information systems, such as sekai-kei (world-type) and iyashikei (healing). In addition, whether an anime belongs to genre A or B is often unclear. As an example, in a previous study (Cho et al. 2017), we defined genre as “categories of anime character-ized by similarities in form, style, or subject matter, including typical genre labels that are universally used by anime users,” but how real users interpreted genre varied. What some us-ers called genre was considered a theme or trope by other users. Even when users used the same genre term, the ele-ments they expected were often different.

In a similar vein, current information systems have shown several limitations in providing anime’s genre infor-mation. While there are a considerable number of user-generated tags about anime genres online, the tags are of-ten confusing and ambiguous, with little consensus for the definitions of many anime genres. Anime and manga tend to use genre terms to indicate the target audience, such as seinen (young men/youth). Some users use the seinen label to describe anime generally made for mature audiences, others use it for anime with complex plots or serious issues regarding life or society, and still others use the term to describe anime with more violent or sexual content. Due to different understandings and ambiguous definitions of the same term, various anime databases commonly show conflicting results when searching for the same genre term. Crunchyroll (http://www.crunchyroll.com/vid-eos/anime/genres) mostly shows anime with mature con-tent under the genre filter “Seinen/Mature” (as the genre term itself already indicates), but Anime-Planet (2017) and MyAnimeList (https://myanimelist.net/anime.php) yield results including Mushi-shi (2005) and Haven’t you heard? I’m Sakamoto (2016), which are both listed in Crunchyroll but not classified as seinen.

Due to definitional ambiguity, it is now common for English-speaking anime users to directly adopt Japanese terms to describe anime genres, such as bishōjo (beautiful girls) and iyashikei. Considering user warrant, literary war-rant, and scholarly usage (Svenonius 1989), information systems may continue to utilize Japanese-originated genre terms. However, using foreign language terms as index terms risks confusing users who do not understand the language and could be problematic for both knowledge or-ganization and information retrieval in the long run.

Additionally, while genre terms are often used across different multimedia, such as comic books, video games, and films, they are not readily usable in anime. Poitras (2002, 34) states that “genre in anime is a tricky subject. Japanese and American animation genres don't exactly par-allel one another; nor are the expressions of genre in Japan quite like those found elsewhere in the world.” Since anime originated in Japan, different cultural notions are interwo-ven in its expressions, including characters, plots, and set-tings of anime. Considering the confusing interpretations, ambiguous meanings, and cultural elements, a more agreed-upon, clearer method for organizing anime genre information is required.

Having agreed-upon definitions of anime genre infor-mation is a vital first step in creating a reliable taxonomy. To fully understand the uses of current anime genre terms and to provide a more robust and clearer way to describe diverse anime genre information, this study explores the following research questions: – RQ1: What are the genre terms of anime that are cur-

rently used, and what kinds of information features are represented in these labels?

– RQ2: What are the facets and foci that describe diverse layers of anime information features currently embed-ded in anime genre terms?

Figure 1. Screenshots of Mushi-shi (2005) and Haven’t you heard? I’m Sakamoto (2016), from left to right.



2.0 Previous studies and practices The current organization of anime genres reveals how in-dividual anime frequently fit into multiple genre terms. Both Poitras (2002) and Richmond (2009) point out that anime itself is not a genre as it is sometimes called. Poitras (2002, 7 emphasis original) states that anime “is in reality an art form that includes all the genres found in cinema, from heroic epics and romances to science fiction and comedy,” while Richmond (2009, v) says that “anime is a medium, not a genre: the cute antics of Hello Kitty and Pokémon, and complex psychological thrillers such as Perfect Blue are all part of the anime universe.” Robbins (2014) and Halsall (2004) also discuss anime as a medium that spans multiple genres.

Poitras (2002) dedicates a chapter to anime genres, dis-cussing several genres and subgenres. The genre/subgenre ordering of the chapter suggests a hierarchical organiza-tion. However, Poitras (2002, 34) points out the problem with assigning only one genre term to an anime:

Gundam 0080 (1989) can be described as a giant ro-bot anime. But such a description does not do justice to the story’s many layers of complexity dealing with family problems, civilians caught up in war, and friendship between a child and soldier, not to men-tion the sheer tragedy of death and destruction

brought on by war. Gundam 0080 is no more just a giant robot anime than War and Peace is just a war novel.

Poitras illustrates the limitations of single identifiers inher-ent in a hierarchical taxonomy. Similarly, Richmond (2009) details genres with some subgenres and indicates a hierar-chical organization. For instance, Richmond includes gen-res like “science fiction” (200), “fantasy and fables” (203), “historical dramas” (217), and “sports, martial arts and contests” (230); some genres are further subdivided, such as “fantasy and fables” into “oriental-influenced,” “Japa-nese folk tales,” “western-influenced,” and “magical girl.”

Moreover, there is a wide variety of disparate labeling systems used by commercial websites. The diversity of browsing systems illustrates a lack of uniform methods of reference for anime and suggests patterns for structuring coherent genre identification (Table 1). Of the five sample websites, four of them use some sort of Boolean opera-tion to allow users to search by multiple genre terms, alt-hough only two, IMDb (http://www.imdb.com) and Crunchyroll, enable a Boolean “and” operation which acknowledges that a single title can encompass multiple genres. IMDb also includes the greatest number of genre terms (twenty-seven), while Amazon lists the fewest (six). Each sampled site provides additional filters to allow users to sort and search by other elements, such as popularity,

Amazon IMDb Crunchyroll Funimation

(https://www. funimation.com/

genre/)

VIZ Media (https://www.viz.

com/watch)

Boolean Operators OR AND AND OR None

Genre Terms

Action; Children’s; Comedy; Fantasy; Horror; Science Fiction

Comedy; Sci-fi; Horror; Romance; Action; Thriller; Drama; Mystery; Crime; Animation; Adventure; Fantasy; Comedy-Romance; Action-Comedy; Superhero

Action; Adventure; Comedy; Drama; Ecchi; Fantasy; Historical; Mecha; Romance; Science Fiction; Seinen/Mature; Shoujo; Shounen; Slice of Life; Sports

Action/Adventure; Comedy; Drama; Fan Service; Fantasy; Horror; Live-Action; Psychological; Sci-Fi; Shoujo; Shounen; Slice of Life

Action-Adventure; Comedy; Drama; Family; Fanasy; Horror; Martial Arts; Mystery; Romance; Science Fiction; Sports; Supernatural; Thriller

Addi-tional Filters

Format; Release date; Franchise; Actor; Director; Customer rating

Popularity; Alphabetical; IMDb rating; IMDb votes; US box office; Runtime; Release date

Alphabetical; Popularity; Updated; Silmulcasts; Seasons

Rating; Version; Date added; Alphabetical; Popularity

Dubbed; Subtitles; Series

Table 1. Examples of current anime information systems.



customer ratings, or release date. Each website includes different search terms and structures its search system dif-ferently, creating unnecessary obstacles for users attempt-ing to locate anime. These search methods acknowledge the multitude of information types that users find desira-ble when either browsing or seeking specific titles. A sys-tem able to efficiently and completely represent a broad variety of information about anime titles in a collection might prove useful to both distributors and users.

The debate on the notion of genre has continued for a long time (Rafferty 2012; Frow 2006; Cobley 2005). Frow (2006) and Cobley (2005) trace the theoretical basis of genre to Aristotle and Plato. Discussing fiction librarian-ship, Rafferty (2012, 557) states that genre “has pragmatic and operative power, but it is a concept that as a categoriz-ing principle has been subject to debate from within liter-ary and cultural theory.” Rafferty (564) also suggests that the social semantic web and user generated tags can be a future solution for developing genres of fiction, arguing that “even within the discourse of literary theory there has been a shift from the idea that genre is based entirely on literary warrant, a viewpoint that underpinned many early approaches to genre theory, to a viewpoint that sees the interaction between the work and the interpreting reader as the basis for meaning-making.” Scholarly tradition has thus begun to lean toward allowing user preferences to shape genre terms and related metadata.

Considering the ambiguous nature of genre, this study applies facet analysis, an analytic approach rooted in the knowledge organization domain (Vickery 1960), to more clearly and rigorously classify anime’s genre information. Facet analysis technique has been used to develop various knowledge organization systems, including classification systems, thesauri, and taxonomies, as well as website archi-tectures and visual/navigational information structures (Shiri 2014). It identifies top-level characteristics of divi-sion (i.e., facets) and allows combination of values (i.e., foci) from within these characteristics of division (Green 2017). As Shiri (2014, 358) noted, facet analysis has proved to be useful to describe the “various characteristics, attrib-utes and aspects of complex, compound and multi-faceted topics.” Since genre has long been understood as both complex and ambiguous, we used a facet analysis approach for this study to explore and collect the different aspects of genre. Without having a comprehensive grasp of all possible aspects of genre, we cannot propose an enumer-ative classification of it. From this perspective, facet anal-ysis provides a lens to view genre as a set of facets, which are “semantically cohesive categories” (Svenonius 2000, 143) and “component parts or ‘the faces’ of a phenome-non” (Smiraglia and Henry 2016, 155).

Further, we adopt facet analysis for its ability to identify different aspects of genre for both indexers and users

(Hjørland 2013). For indexers, the faceted schema pro-vides a structure for describing anime. By following the structure, an indexer is less likely to miss important facets. For users, the list of facets presents possible access points of an information system and helps them formulate que-ries. In addition, as Green (2017) points out, the main ad-vantages of facet analysis are its strong link to a field’s lit-erature and its ability to address organizational needs. Be-cause our data sources are information systems that em-phasize anime retrieval and discovery, and because the pri-mary objective of this study is to contribute to the im-provement of both anime description and retrieval, we be-lieve that the facet analysis approach is appropriate.

Following the basic steps for facet analysis (La Barre 2010), we explore multiple components of anime genre. Starting from collecting genre terms from information sys-tems for anime, we develop facets and sub-facets after data cleaning and organizing the foci, and all the concepts de-rived from the analysis were mutually exclusive (Slavic and Davies 2017). Definitions of all facets and some of the foci terms are described in section 4. Findings and Discus-sion. 3.0 Research method As emphasized in the literature review, anime genre is complex and multifaceted. There is no consensus for its definition. Further, genre is often used interchangeably with terms like theme and plot. The distinctions between these terms are ambiguous. The study’s purpose is to ex-amine the concept of genre in anime and identify promi-nent aspects. Hence, we applied facet analysis to anime genre terms collected from 36 English websites (3 addi-tional Japanese websites were observed to see if there is any important information missing from English websites, but they were not part of the analysis). Applying an induc-tive approach, we identified nine major facets from 1,597 terms using a card-sort method. 3.1 Data collection The data collection process is visualized in Figure 2.

We collected anime genre terms from popular websites and blogs. Sources include information systems providing anime information, such as Anime-Planet. These systems allow users to browse and search anime by genre. Other sources list or define anime genre terms, such as Wikipedia and Senpai Knows.

The criteria for source selection are language and term variety. To understand anime genre terms in the context of English-speaking audiences, the language criterion set the focus of this study at sources which provide English-lan-guage anime genre terms. As a preliminary attempt to



compare English and Japanese anime genre terms, besides the thirty-six English language sources, we examined three Japanese sources for reference (i.e., Kinokuniya web store, Tsutaya web store, and d Anime store), all of which are major online stores selling anime DVDs and Blu Rays or are streaming providers. Another criterion is term variety. While identifying prominent facets of anime genre through a bottom-up approach, we selected sources with a variety of anime genre terms. There is a range of seven to 283 genre terms, with subdivided terms in each source se-lected.

The principles for term collection are transcription and context preservation. For sources that organize genres hier-archically, we collected the terms of each subdivision and preserved the context. For instance, both “sports” and “game” are top level genre terms; and both terms have the subdivision term “baseball.” In this case, we collected four terms: “sports;” “sports—baseball;” “game;” and, “game—baseball.” Documenting the hierarchy in this way helps pre-serve the context and meaning of a subdivided term. We collected 1,597 terms from thirty-six English-language sources.

3.2 Data cleaning There are five stages of data cleaning. The first stage is merging duplicates, in which identical terms are merged. After the first stage, we made the terms into cards and con-ducted card sorting for the second and third stages. The second stage is merging variant forms. Variant forms of the same term may differ in spacing, usage of hyphen, and Romanization (e.g., lolicon and rolikon [Lolita complex]). Obvious typos are also merged at this stage. The third stage is merging translations and romanizations. For in-stance, we merged magical girl with mahou shoujou. After the first three stages of data cleaning, we sorted the terms into three groups: simple terms, compound terms, and sub-genre terms. Simple terms describe one concept (e.g., mys-tery, love triangle, mind-twisting, piloted robot). Com-pound terms address more than one concept (e.g, ac-tion/adventure, military and warfare, anime and manga). Sub-genre terms refer to terms including hierarchical structure. For instance, “game—baseball” specifies that the sub-genre term “baseball” is subdivided under “game.” Considering the complexity of the three types of terms,

Figure 2. Data collection process.



we used simple terms as the foundation for facet develop-ment.

The fourth stage is merging similar simple terms with dif-ferent levels of specificity. For instance, we grouped after-life, underworld, heaven, hell, and makai (world of spirits; hell) and labeled them as “afterlife.” In this case, “afterlife” was selected as the preferred term to represent this group of similar terms. Another example is using “game” to rep-resent different types of games, including “board games,” “online computer gaming,” “high stakes game,” and “action game.” In this stage, we merged similar or closely related terms with different levels of specificity and labeled the group with a broader term. The broader term is the pre-ferred term encompassing the concepts of the group of terms. The data cleaning in this stage adjusts the granularity of the data. Instead of focusing on the different meanings of similar or closely related terms, this approach directs our focus to identify prominent general concepts about anime genre. The fourth stage thus establishes the foundation of facet analysis for this study.

The fifth stage is standardization. We applied three prin-ciples for selecting and standardizing preferred terms. The first principle focuses on the part of speech and singular or plural forms. Preferred terms are terms with the highest fre-quency. If there is no prominent preference identified, we use plural nouns. The second principle standardizes the rep-resentation for romanizations and English terms. For terms having both Japanese romanization and English translation, we use romanization and place the English term in paren-thesis. An example is “kodomo (children).” The third prin-ciple focuses on disambiguation. For terms encompassing multiple concepts, we use parenthetical qualifiers to disam-biguate. For instance, the concept “gekiga” refers to a com-bination of a type of plot and a realistic visual style. We use “gekiga (Plot)” and “gekiga (style)” to differentiate the spe-cific aspect of focus.

The five stages of data cleaning render a total of 643 terms, including 538 single and compound terms and 105 sub-genre terms. 3.3 Data analysis There are two phases in the data analysis process. The first phase develops facets and sub-facets using a bottom-up approach. We examined the preferred terms that appeared more than once after data cleaning. These terms are the foci. Based on the foci, we developed nine facets and thir-teen sub-facets. After the first phase, each focus was as-signed to one facet and one facet only. Foci assignment tests the mutual exclusivity of the facets.

The second phase tests the exhaustivity of the facets with compound terms and sub-genre terms. As defined in the data cleaning section, compound terms and sub-genre

terms include more than one concept. We used the facets and foci to describe all concepts presented in the genre terms collected and adjusted terms accordingly.

We recognize that new genre terms continue to emerge and that the genre terms collected are not comprehensive. However, the purpose of this study is not to provide an exhaustive list of genre terms. Instead, we propose a fac-eted structure to organize anime genre descriptors. Through collecting various terms, the goal is to identify prominent dimensions of anime genre. We present the re-sults of this study by developing mutually exclusive and collectively comprehensive facets that are flexible enough for expansion. Further, we collect existing terms that users already apply for anime information, organization, and re-trieval. Some terms present challenges and reaffirm the complex nature of genre. We will discuss some of these challenges below. 4.0 Findings and Discussion 4.1 Overview of Developed Facets We identified nine facets and 153 foci terms. Each facet rep-resents a different aspect of anime, mutually exclusive from other facets. The facets, the number of foci within each facet, and the examples of foci are shown in Table 2. The complete list of subject foci terms is presented in an appen-dix.

Some of the facets, such as “setting,” “plot/narrative,” “subjects,” “feature” and “production” have sub-facets to represent specific hierarchical relationships. For example, “setting” has three sub-facets: “setting-time,” “setting-situa-tional,” and “setting-place.” All the foci of “setting” belong to one of these sub-facets. In the case of “plot/narrative,” most of its foci directly belong to the main facet, “plot/nar-rative,” but one of its foci, “romance,” is further divided into “romance-types” and “romance-multiple people.”

It was challenging to construct facets that: 1) distinguish clearly among the component parts; 2) are relevant to the purpose, subject, and scope of the

classification; 3) represent characteristics of division that can be meas-

ured; 4) only represent single characteristics of division; 5) are mutually exclusive to each other; and, 6) are derived based on the nature of the subject being

classified (Spiteri 1998). Though we are satisfied with the facets presented in this study, we acknowledge that classifying subjective infor-mation of a cultural medium will evolve over time. To ac-commodate the dynamic nature of anime genre terms, the



proposed facets are flexible. New genre terms may be added to the faceted schema, and the structure can be ex-tended as well.

Note that some of the foci are written in Japanese and have a translated meaning inside of the parentheses beside the original Japanese term, such as “josei (women)” and “iyashikei (healing).” As discussed above, it is very com-mon in the current US anime community for anime fans to use Japanese terms to describe genre information. Con-sidering user warrant (Svenonius 1989), we could have adopted the current term practices directly into our faceted schema. However, for the faceted schema to be more in-tuitive and user-friendly to people who are not familiar with anime and Japanese, we decided to have a translated meaning together with the commonly used Japanese genre term to increase ease of use for English language indexing. 4.2 Audience The Audience facet encompasses the target demographic of the anime. Thus, the foci terms tend to originate from the creators, publishers, and/or distributors of anime. The rea-son for including the target audience as a facet is twofold.

First, it draws on existing categories for organizing anime. On Crunchyroll, for example, shōnen, shōjo, and seinen are included as genres. Anime-Planet and MyAnimeList use shōnen, shōjo, seinen, and josei, with MyAnimeList also provid-ing “kids” as a genre term. Denison (2015, 24) points out that “anime is for everyone in Japan, but not every show is aimed at everyone—there are genres for most age groups, genders and sexual orientations, and anime’s genres often reflect industrial categories” and that the terms most fre-quently applied to anime in Japan come from the manga de-scriptors for intended audiences. Denison adds that “in Ja-pan, shōnen and shōjo are themselves, therefore, sprawling transmedia categories whose meanings can seem confusing or seemingly arbitrary.” So, while there is strong literary and user warrant for the “audience” facet’s foci terms, the boundaries between the categories they pull from are not always clear.

Second, the “audience” facet highlights that the in-tended demographic of an anime can have an impact on the content. That is, there are various tropes associated with each of the foci. Shōnen anime often incorporate fighting and male protagonists, while shōjo anime often fea-ture romantic subplots and female protagonists. However,

Facet Number of

Foci Examples of Foci

Audience 6 Josei (Women), Seinen (Young Men/Youth), Shōnen (Boys), Shōjo (Girls), Kodomo (Children), Family

Setting 15

- Setting-Time: Cyberpunk, Steampunk, Historical - Setting-Situational: Fantasy, Apocalypse, Science Fiction, Science Fantasy, Dystopia - Setting-Place: Isekai (Parallel Universe), Western, Restaurants/Stores, Space Opera,

Space, School, Afterlife

Mood 7 Tragedy, Thriller, Comedy, Psychological, Horror, Suspense, Iyashikei (Healing)

Character 32 Detectives, Police, Students, Idols, Pirates, Hunters, Hybrids, Cyborgs, Androids, Ro-bots/Mecha, Bishōjo (Beautiful Girls), Bishōnen (Beautiful Boys), Monsters, Aliens, Vampires, etc.

Plot/Narrative 21

Slice of Life, Coming-of-Age, Adventure, Mystery, Absurd, Romance, etc. - Plot/Narrative-Romance-Types: Bara (Boys’ Love for Male Audience), Age Gap,

Shōnen Ai/Yaoi (Boys' Love), Shōjo Ai/Yuri (Girls' Love) - Plot/Narrative-Romance-Multiple People: Love Triangle, Harem/Reverse Harem

Subjects 47 Game, Vehicles, Cooking, Time Travel, Magic, Virtual Reality, Politics, Work Life,

Crime, Arts, Natural Disaster, Sports, etc. - Subjects-Sports: Wrestling, Baseball, Gymnastics, Boxing, Martial Arts, Archery, etc.

Association with Other Works 5 Parody, Novel, Comics, Manga, Documentary

Feature 7 Action Scenes, Fan Service, Power Suits, Henshin (Transformation) - Feature-Action Scenes-Types: Proxy Battles, Swordplay, Gunfights

Production 13

- Production-Techniques: CG Animation - Production-Styles: Chibi (Super Deformed), Gekiga (Style) - Production-Regions: Foreign Anime, Cartoon - Production-Agents: Voice Actors - Production-Formats: ONA (Original Net Animation), Special (SP), OVA, Short

(Length), Movies, Series - Production-Languages: Dubbed

Table 2. Anime genre facets with examples of genre terms.



it is important to note that the actual demographic watch-ing anime can differ from the intended demographic. – Kodomo (Children) is for anime intended for children

with an upper bound of late elementary school age to junior high school age;

– Family is for anime intended for all ages based on a hy-pothetical family containing young children to adults/elderly;

– Shōnen (Boys) is for anime intended for teenage boys; – Shōjo (Girls) is for anime intended for teenage girls; – Seinen (Young Men/Youth) is for anime intended for

young men; and, – Josei (Women) is for anime intended for women. Because “kodomo” and “family” anime both have young children as all or part of the intended demographic, the boundary between them is difficult to determine. Richmond (2009, 201) calls My Neighbor Totoro and Doraemon anime for children in his discussion of anime demographics. However, IMDb lists the genres of My Neighbor Totoro as “animation, family, fantasy” and those of Doraemon as “animation, com-edy, family, sci-fi.”. The Japanese Wikipedia entries for the 1979 (Wikimedia 2017a) and 2005 (2017b) Doraemon anime further complicate matters by listing both as falling under the “family/intended for general public anime” (famiri ippan-muke anime けアニメ一般向ファミリー・ ) and “science fiction gag anime” (SF gyagu anime SFギャグアニメ) gen-res. One final wrinkle is that the Japanese Wikipedia page for “intended for children anime” (Kodomomuke anime

けアニメ子供向 ) links to the 2005 Doraemon anime from its section on the characteristics of “Kodomo” anime for infants and young children (2017b). If the creator, publisher, or distributor of an anime that could fall into either “kodomo” or “family” does not provide the intended de-mographic, then assigning either “kodomo” or “family” to an anime becomes a subjective decision based on whether the cataloger feels the anime is strictly for young children.

Seinen (青年) as a term can mean “youth,” though typi-cally it is used to refer to young men. Denison (2015, 24) illustrates the difficulty of the term by translating it two dif-ferent ways on the same page: first as “young adult” and later as “aimed at adult men.” It may be the case that some of the sites are using the term in the broader sense of referring to all youth regardless of gender. However, Richmond (2009), Brenner (2007), and Poitras ([2007] 2010) use the term in the narrower sense of young men, which is supported by several Japanese dictionaries (Kodansha’s 1999; Kodansha’s 2002; Breen 2018). Another difficulty with seinen arises be-cause most English-language sites do not provide the kanji for Japanese terms. Seinen is also the pronunciation of 成年, meaning “adult age; majority,” 盛年, meaning “prime of life,” and 生年, meaning “year of birth; age.” While the last

two phrases are unlikely to be what was meant on the vari-ous sites harvested for terms, it is possible that some of the sites meant the word meaning “adult age” as a means of referring to adult content. The comparatively mature con-tent of many seinen anime also lends itself to this possibility. 4.3 Setting The “setting” facet describes the environment the anime takes place in. The facet is split into three sub-facets, which represent the different aspects of setting: “setting-time,” “setting-place,” and “setting-situational.”

The “setting-time” sub-facet contains terms that relate to the temporal aspect of setting. As such, it situates the anime in time if the anime takes place on earth or provides an equivalent time period if the anime takes place in an alternate world: – “Historical” is for anime that take place during a past

time period; – “Steampunk” is for anime that take place in a time pe-

riod where steam power is prevalent; and, – “Cyberpunk” is for anime that take place in a future

time period where advanced technology intersects an abnormal social order.

The “setting-place” sub-facet contains terms that relate to the spatial aspect of setting. As such, it provides the loca-tion the anime takes place in: – “Isekai (parallel universe)” is for anime that take place

in an alternate reality; – “Western” is for anime that take place in the Western

United States; and, – “Space opera” is for anime that take place across an

extensive expanse of interstellar space. The “setting-situational” sub-facet contains terms relating to an abnormal state in the setting’s place, time, or other characteristics. It provides a description of the aesthetic or other circumstances of the anime: – “Fantasy” is for anime that are set in a world where

some level of magic or the supernatural is present; – “Apocalypse” is for anime that are set in a world af-

fected by cataclysmic events; – “Science fiction” is for anime set in a world where sci-

ence and advanced technology play a role; and, – “Dystopia” is for anime that are set in a society or

world where living conditions are poor. The “setting-time” sub-facet currently lacks a robust set of foci to describe when an anime is set. “Historical” as a



term is rather broad, which limits its usefulness in search-ing for anime. More granularity in descriptors of the past would facilitate locating desired works. The hazards in breaking down the “historical” label too far are that cata-logers may have difficulty identifying the proper label and searchers may locate only a small number of anime. Terms associated with “historical” must be broad enough to fa-cilitate easy cataloging and narrow enough to communi-cate useful information in search results. A further diffi-culty will include choosing terms that encompass time-based information without also necessitating other de-scriptors; for example, while the term “medieval” would clearly indicate a relatively narrow set of centuries, it would also presuppose a “setting-place” facet of Europe.

“Cyberpunk,” “steampunk,” “western,” “fantasy,” and “science fiction” are all problematic because the terms have elements that fit into more than one setting sub-facet. While “cyberpunk” anime are usually set in a future “set-ting-time,” the word also requires “setting-situational” as-pects in terms of technological and societal characteristics. In other words, “cyberpunk” is not exclusively an indicator of time. Similarly, a “western” is often conceived of as be-ing situated in the Western United States (“setting-place”) in the second half of the nineteenth century (“setting-time”) with certain conventions (“setting-situational”). However, as Tudor (1974, 139) points out, “there is no ba-sis for assuming that a ‘Western’ will be conceived in the same way in every culture. The way in which the genre term is applied can quite conceivably vary from case to case.” Cowboy Bebop draws from the “western” genre but is not set in the time and place of most Westerns. “Fantasy” and “science fiction” proved especially difficult to pin down as the research team’s members have varied under-standings of what those genres entail.

The “setting-place” sub-facet is challenging because of the variety of places available; some anime span galaxies or planets whereas other anime can be confined to a single location, such as a school. Some anime also begin in one location, but then move out to other locations. For in-stance, Space Patrol Luluco could potentially be said to fall under the scope of both the “school” and “space” terms. Because of this it will be necessary to determine whether only the most prevalent setting should be listed for a par-ticular work or whether all applicable settings should be considered. 4.4 Mood According to Turco (1999, 50 emphasis original), “Atmos-phere is the mood of the narrative, and mood is created by means of setting (locale and surroundings in which the narrative takes place), attitude (of the narrator and of the characters in the narrative), and descriptions.” In media stud-

ies, mood can have two different contexts: certain atmos-pheres represented by a particular work (in this case, anime), or the emotional states of the audience. For ex-ample, one might want to watch an anime that has a tragic mood (former) to alleviate his or her current sad feeling (latter). In this study, mood is the emotional state of the audience in relation to the feelings an anime elicits.

While the definition suggests using adjectives for the foci terms, we use the most common forms from the data collected. Most “mood” terms are nouns when used as genre terms: – “Psychological:” An anime emphasizing the mind, fo-

cusing on the motivations or mental states of the char-acters. A psychological anime often surfaces questions which initiate thought-provoking, agitated, unsettling, or disturbing feelings.

– “Iyashikei (healing):” Iyashikei means “healing” in Jap-anese. The audience feels healing and soothing. Many “iyashikei (healing)” anime have a calm and slow-paced atmosphere, depicting small delights in daily life, a ten-dency which overlaps with the “slice of life” focus un-der the “plot/narrative” facet.

An anime may trigger different moods, and through facet analysis, we can describe different moods presented in a single anime by assigning multiple foci terms. As presented in the “iyashikei (healing)” example above, the “mood” terms are often closely related to “plot/narrative” terms. Audiences often feel “healed” by watching anime with a “slice of life” plot. 4.5 Character “Character” describes the figures in an anime. It covers hu-man, non-human, and imaginary figures. There are terms referring to characters of specific occupations (e.g., “pi-rates,” “maids,” “students”) and those with specific charac-teristics (e.g., “bishōjo [beautiful girls],” “rorikon [Lolita complex]”). Some terms describe imaginary creations as characters, such as “robots/mecha,” “demons,” and “vam-pires.”

As mentioned in the research method section, consid-ering the granularity of terms, some foci terms are used for concepts that are not identical but similar. For instance, “gangs” is used for both Mafia and Yakuza (crime groups that originated in Japan); and “monsters” is used for yōkai (supernatural creatures in Japanese folklore). Depending on local needs and collections, an information system may apply this faceted structure and decide to merge or divide similar terms. The foci terms generated from this study are not an enumerative list. The faceted structure is flexible for expansion and provides a customizable level of specificity.



The adjustment of specificity of foci terms does not affect the nine facets identified. 4.6 Plot/narrative “Plot/narrative” is defined as how the story unfolds. In lit-erary discourses, Brooks (1984, XI) defines plot as “the de-sign and intention of narrative, what shapes a story and gives it a certain direction or intent of meaning.” Similarly, Turco (1999, 101) defines it as the “series of rising actions leading to the climax in a written or theatrical narrative; the thread of associated incidents that makes a story, including its beginning, middle, and end” (emphasis original). In other words, plot occurs when narratives, or accounts of events (Wolfreys, Robbins, and Womack 2016), are ordered in a cer-tain way with a certain intention. In this regard, this facet is distinguished from “subjects” in that “plot/narrative” fo-cuses more on the overall development of the storyline, in-stead of overarching themes or topics. Among the twenty-one foci terms, there are six sub-terms under “romance.” Some of the foci terms in this facet are: – “Romance:” The main body of the story deals with

love; – “Romance-types:” Different types of romance; – “Bara (boys’ love for male audience):” “Romance”

anime in which the main characters are both male; often created by male artists, for a male audience;

– “Shōnen ai/yaoi (boys' love):” “Romance” anime in which the main characters are both male; generally tar-geted towards women;

– “Shōjo ai/yuri (girls' love):” “Romance” anime in which the main characters are both female;

– “Age gap:” Anime that shows a romantic relationship between two main characters with a significant age gap;

– “Slice of life:” A story that shows the everyday life of the characters;

– “Coming-of-age:” A story that depicts the process by which young characters grow up, by showcasing char-acters who make the transition naturally or who are forced to grow due to external motivations;

– “Sekai-kei:” As a neologism which has not been clearly defined, this term typically refers to a story that shows an association between two main characters’ relation-ship and a crisis in the world, e.g., Neon Genesis Evangelion (1995);

– “Absurd:” A story that is wildly unreasonable, illogical, or inappropriate; and,

– “Gekiga (plot):” A story that is likely to be serious and mature, with realistic and darker settings.

Among these foci terms, “sekai-kei” and “gekiga (plot)” re-quire more discussion. Unlike other foci in this faceted

schema, “sekai-kei” does not have any translated meaning next to the term. The reason is that the definition of this term, “sekai-kei” itself, has not been clearly agreed upon or developed by anime communities. Thomas (2014, ii) states, “The term sekai-kei (“world-type”), a genre of subcultural narrative whose defining characteristic is the omission of a mediating social dimension between individual characters and an apocalyptic large-scale situation, has come to be used as a tool of cultural criticism pertaining to young adults of the mid-90s to the mid-2000s.” Satoshi Maejima’s book Sekai-kei to wa nani ka: Posuto-Eva no otaku shi further explains “sekai-kei.” Maejima (translated and paraphrased by Thomas 2014, 14-15) finds that “sekai-kei” anime share some characteristics: “1. They omit the social setting, 2. What is fought and on which principle it functions is not understood, and 3. These omissions are deliberately in-cluded in order to induce sympathy for the protagonists’ self-consciousness or the (often tragic) love between the protagonist and the heroine in the consumers” (emphasis original).

Other than the somewhat forced and unnatural transla-tion “world-type,” which does not mean anything specific in English, there is not a good translation to define what “sekai-kei” indicates. The fact that there is not a clear-cut consensus on what the genre means does not help the situ-ation, either. The research on the term and discussions fi-nally lead us to decide to keep the term as it is, “sekai-kei,” a decision which may indicate a limitation of the faceted schema.

Another foci-term that needs to be discussed further is “gekiga (plot).” As a term that is often used to describe cer-tain darker genres of manga and anime, “gekiga” originated in “the late 1960s, when new styles and approaches to con-tent were explored by a variety of manga artists” (Brophy 2010, 130). Brophy (131) goes on to describe how “gekiga” is expressed in some of Osamu Tezuka’s works:

If the earlier works invited identification with the cud-dlesome, the gekiga works portend a harsh duality in their theatricalized two-facedness. Centered as so many of the stories are on the duplicitous and deceit-ful maneuvers of humans, the postcute effect of their visages serves to amplify the psychological mecha-nisms that motivate their actions. From portraits of self-disorienting quandaries of personal identity (Ru-dovihi B, Apporo no uta, and Bomba!) to investigations into self-destructive disregard for personal ethics (Tsumi to batsu [Crime and Punishment], Mu, Ningen konchu ki) to treatises on self-revelatory moments of personal enlignment (Kirihito sanka, Buddha, Hi no tori), Tezuka’s characters teeter on the brink of redefinition and reincarnation (emphasis original).



The tricky part of “gekiga” is that the term indicates two different aspects of anime, the story and the visual style. Generally, “gekiga” anime can be identified by their focus on realistic artwork and character design and their inclu-sion of mature and serious stories. Thus, placing the foci under a single facet does not satisfy the condition of being mutually exclusive, because the schema contains both “plot/narrative” and “production-styles.” Therefore, we created two separate terms, “gekiga (plot)” and “gekiga (style)” to differentiate the nuances and place them sepa-rately under two different facets. 4.7 Subjects The “subjects” facet refers to themes or topics an anime pri-marily conveys. The facet contains thirty-three foci terms, including “game,” “cooking,” “time travel,” “magic,” “tour-naments,” “politics,” “work life,” “crime,” “arts,” “super-powers,” etc. One foci-term from the facet, “sports,” has fourteen child terms, including “wrestling,” “baseball,” “gymnastics,” “archery,” and “ice skating,” indicating the popularity of “sport”s subjects in anime. Although the foci terms listed here under “subjects” are not the complete set of topics existing in anime, they are the subjects that ap-peared more than one time in the databases researched. When there are more “subjects” terms retrieved in the fu-ture, they can be easily added to the current structure. 4.8 Association with other works “Association with other works” describes a facet that defines connections to works that are largely referential in nature. For example, an anime falling under the facet may use a cli-ché gag or reference a famous manga or other entertainment work (situating historical work, as in Mushi-shi), displaying cultural connectivity between creative works. These connec-tions often occur due to the use of manga as a basis for anime stories and production; however, it is also the case that anime can be used to display a certain rhetorical under-standing of Japanese culture. Foci terms included are “par-ody,” “novel,” “comics,” “manga,” and “documentary.” 4.9 Feature Feature describes a facet that is highly specific yet increas-ingly general and regularly presented within anime. In a sense, these are clichés that have become descriptive of works unto themselves, such that they are, at times, the en-tirety of the work, rather than specific instances within the work. Still, there are also many anime where these aspects are simply inserts, presented in a narrow manner for a cer-tain audience, with little consequence or meaning with re-gards to the rest of the show:

– “Fan service:” work that contains “service”; typically, some form of sexualization of male or female charac-ters, largely for the audience rather than the show;

– “Power suits:” Suits or other forms of mobile armor that provide characters extra power in some form—in-cludes Gundam-type armors, but also exoskeletons and other forms of armor enhancement; and,

– “Henshin (transformation):” scenes that show off a character changing form in some way, typified by magical girl series, though before this, often included in Power Rangers and other series involving characters changing to a “powered up” form.

4.10 Production “Production” represents a wide variety of terms for how the anime was created. This includes the different ways in which an anime may be brought about by way of presen-tation. The facet relates strongly to style visually as well as acoustically and includes a wider discussion of how length defines the varying types of anime typically created. Some of the sub-facets and foci terms are as follows: – “Techniques∷ the method for constructing the anime

visually; – “CG animation:” a visual method using digital com-

puter graphics to render characters or environments, ra-ther than hand-drawn animation;

– “Styles:” the normative visual representation of charac-ters or backgrounds in the anime;

– “Chibi (super deformed):” a style defined by characters with cartoonishly large heads and much smaller bodies, often focusing on representational facial animations;

– “Gekiga (style):” a style defined by dramatic imagery, including characters in relief, often including wide pan-oramic shots and specific use of shadow to denote fa-cial features;

– “Regions:” the location in which the anime is made; – “Foreign anime:” anime developed outside of Japan or

of the normative culture defining it; – “Cartoon:” shows developed or produced with west-

ern audiences in mind, typically more stylistic than spe-cific to location;

– “Agents:” groups or individuals involved with the pro-duction of anime;

– “Voice actors:” individuals who provide audio for char-acters in anime;

– “Formats:” the structure of the anime either as it re-lates to length or connectedness to other works;

– “ONA (original net animation):” a typically short work produced for immediate release online, rather than other methods of broadcast;



– “Special (SP):” a work that is typically a one-time event, presented for a specific purpose, such as a Christmas special;

– “OVA (original video animation):” a short work pro-duced for immediate release to some fixed physical for-mat, such as VHS, Blu-ray, DVD, or other forms of fixed media.

– “Short (length):” a short is defined both by length and presentation, typically in a trailer-like fashion, fre-quently presenting or expanding upon some quirk or oddity in a work;

– “Movies:” longer works presented in a non-serialized manner, often complete and standalone works, without need for connective tissue from other works or epi-sodes; and,

– “Series:” shorter, episodic works that are connected by a consistent cast of characters, storylines and events, often culminating or building up over time.

4.11 Facet application to existing anime To show how our schema can describe different genre in-formation in anime, Table 3 shows facet and foci applica-tion to existing anime as examples. Note that multiple foci terms can be applied to each facet due to their mutual ex-clusivity, which makes the genre facet classification stronger by enabling users to describe anime in detail. When a specific facet of an anime is ambiguous, unclear, or cannot be described well with the current foci, it is left with N/A.

One Punch Man (2015)

Attack on Titan (2013)

Ristorante Paradiso (2009)

Ouran High School Host Club (2006) One Piece (1999)

Audience Seinen (Youth) Seinen (Youth) Josei (Women) Shōjo (Girls) Shōnen (Boys)

Setting Science Fiction Situational: Apocalypse

Place: Restaurants/Stores

Place: School Situational: Fantasy

Mood Comedy

Tragedy; Thriller; Horror;

Suspense

Iyashikei (Healing) Comedy Comedy

Character

Hero (Protagonist);

Cyborgs; Monsters;

Aliens; Samurai;

Ninja

Hero (Protagonist);

Monsters; Hunters

Heroine (Protagonist)

Heroine (Protagonist); Students;

Bishōnen (Beautiful Boys)

Hero (Protagonist);

Pirates; Hybrids

Plot/ Narrative

Coming-of-Age; Adventure

Coming-of-Age; Gekiga (Plot);

Survival

Romance; Romance-Types: Age

Gap; Slice of Life;

Drama

Romance; Romance-Multiple People: Harem/

Reverse Harem; Slice of Life

Coming-of-Age; Drama;

Adventure

Subjects Superpowers; Violent Contents

Violent Contents; War;

Military;

Cooking; Work Life

Club Superpowers;

Violent Contents

Association with Other

Works Manga Manga Manga Manga Manga

Feature Action Scenes Action Scenes;

Swordplay Fan Service Fan Service Action Scenes

Production Formats: Series; OVA (Original

Video Animation)

Formats: Series; OVA (Original

Video Animation);

Styles: Gekiga (Style)

Formats: Series Formats: Series

Formats: Series; Movies;

OVA (Original Video Animation)

Table 3. Facet application to existing anime examples.



5.0 Challenges and Implications The biggest challenge confronted was the choice of foci terms for each facet. Because anime naturally adopts Japa-nese culture and language to describe its genres, most Eng-lish-speaking users also tend to use Japanese terms to de-scribe anime’s genres, which has become communally nor-mative. Considering user warrant and literary warrant (Svenonius 1989), we believe the genre terms currently used by real users should be reflected in our schema for users who are already familiar with the material. Using ex-isting terms also demonstrates respect for the culture, do-main, and community.

At the same time, as we briefly discussed in the Intro-duction section, adopting foreign terms to the organiza-tion system thoughtlessly, especially when there are many foreign terms, may yield several serious issues. The organ-ization system should be user-friendly to new or casual us-ers. That is, the terms themselves used in the schema should be intuitive enough so users would not require much training to learn how to use the schema. With these conflicting goals in mind—honoring existing terms and cultures and providing an intuitive schema—we discussed three different options for term selection: using original (Japanese) terms, using English-translated terms, and using both together.

Using English-translated terms might create a more in-tuitive schema, especially for users who are not familiar with anime terminology. However, it would not be “friendly” to anime fans and people who are familiar with the culture already, ironically. Additionally, several terms could not be translated easily because of their complex meanings or interrelationships with either Japanese or anime culture. As examples, we discussed “gekiga” and “sekai-kei” above. The prevalence of difficult-to-translate terms is the reason we chose to use both English and Jap-anese, having original terms and translated terms together in our schema. Using such a method, we can keep the orig-inal terms used in the community and have a schema intu-itive enough for users unfamiliar with anime jargon.

Any researchers in the knowledge organization domain will likely face linguistic issues when they attempt to build a new schema for relatively new types of information or foreign materials, including art, design, visual media, and more. We hope our approach might provide some guide-lines for future research. 6.0 Conclusion By harvesting genre terms from a diverse spread of online anime collections and investigating the relationships be-tween them, we identified common types of information represented in related terms. These commonalities were

condensed into general facets of genre, listing specific op-tions as foci terms under each facet. By organizing genre terms in such a manner, our system allows each anime title to encompass multiple genres.

The faceted approach to anime genre, furthermore, re-veals connections between genre terms in a way that is not necessarily visible from traditional organizational methods for genre. Traditional hierarchical methods limit genre ele-ments to related concepts, such as broader terms or nar-rower terms, while our analysis reveals that genre includes more aspects than the traditionally narrow labels for gen-res and sub-genres. Using facets to address various aspects of a work (setting, mood, character types, etc.) demon-strates how more than one type of information can be rep-resented in a genre organization system. Instead of re-stricting an anime’s genre to one choice in a hierarchical system, a structure that provides multiple labels allows us-ers to accurately represent the multi-genre nature of anime. For example, while a story that includes comedic and romantic elements might traditionally be labeled a “ro-mantic comedy,” as its own genre, that story could in our faceted approach support both a mood genre of comedy and a narrative/plot genre of romance, more clearly iden-tifying the work’s characteristics.

In addition, the faceted approach is easily applicable to formats beyond anime. Genre as a universal concept can be useful and flexible when describing existing works. Granted, multifaceted genre labels have limited usefulness within the physical space of a library or other physical col-lection, in that a single DVD cannot be shelved in multiple places at the same time. But, a multifaceted genre system has complete freedom within modern electronic indexing and searching systems to make works accessible to users. No matter how many different search terms or facets a user includes in a search, the database can still point the user to relevant results. A faceted system such as we pro-pose is also easier to extend than traditional hierarchies. While a hierarchical system must flex in many directions to accommodate the addition or deletion of a term by es-tablishing new pathways between terms, a faceted system can easily absorb new terms within various facets.

We found disparate conceptions of genre and a lack of uniformity among anime search systems restrictive to the usefulness of genre data. For users, breadth and specificity of data are desirable. A faceted system allows anime to be identified by breadth of information (in listing multiple as-pects, such as setting, theme, and target audience) and by specificity of information (in selecting as many labels as necessary from within each aspect). Instead of a traditional hierarchy, a faceted approach better organizes and repre-sents the nebulous conceptual space of genre; a faceted approach welcomes the complexity of anime works rather than limiting them to single spaces in a genre catalog.



References AJA (Association of Japanese Animations). 2016. “Anime

Industry Report 2015.” http://aja.gr.jp/english/japan-anime-data

Amazon Services LLC. 2017. “Amazon.com: Movies & TV: Anime.” Accessed July 25. https://www.amazon. com/anime-dvd-bluray/b/ref=MoviesHPBB_Genres

_Anime?ie=UTF8&node=2650364011 Anime-Planet (blog). 2017. “Complete List of Anime Gen-

res and Tags.” Accessed July 25. https://www.anime-planet.com/anime/tags

Breen, Jim. s.v. “青年 ,” Accessed July 25, 2017. “Jim Breen’s WWWJDIC.” [Melbourne}: Monash Univer-sity, Electronic Dictionary Research and Development Group. http://nihongo.monash.edu/cgi-bin/wwwjdic ?1MDJ%C0%C4%C7%AF

Brenner, Robin E. 2007. Understanding Manga and Anime. Westport, CT: Libraries Unlimited.

Brophy, Philip. 2010. “Osamu Tezuka’s Gekiga: Behind the Mask of Manga.” In Manga: An Anthology of Global and Cultural Perspectives, ed. Toni Johnson-Woods. New York: Continuum, 128-36.

Brooks, Peter. 1992. Reading for the Plot: Design and Intention in Narrative. Cambridge, MA: Harvard University Press.

Cho, Hyerim, Marc L. Schmalz, Stephen A. Keating and Jin Ha Lee. 2017. “Information Needs for Anime Recom-mendation: Analyzing Anime Users’ Online Forum Que-ries.” In JCDL '17 Proceedings of the 17th ACM/IEEE Joint Conference on Digital Libraries 305-6 Toronto, Ontario, Can-ada—June 19-23, 2017. Piscataway, NJ: IEEE Press.

Cobley, Paul. 2005. “Objectivity and Immanence in Genre Theory.” Genre Matters: Essays in Theory and Criticism, ed. Garin Dowd, Lesley Stevenson & Jeremy Strong. Bris-tol: Intellect, 41-53.

“Crunchyroll – Browse Anime by Genres.” 2017. Crunchy-roll (blog). Accessed July 20. http://www.crunchyroll. com/videos/anime/genres

Davis, Northrop. 2016. Manga and Anime Go to Hollywood. New York: Bloomsbury Academic.

Denison, Rayna. 2015. Anime: A Critical Introduction. Lon-don: Bloomsbury Academic.

Exner, Nina. 2012. “Anime-zing in North Carolina: Li-brary Views of Anime Fans.” North Carolina Libraries 70, no. 1: 28-34.

Fennell, Dana, Ana S. Q. Liberato, Bridget Hayden, and Yuko Fujino. 2013. “Consuming Anime.” Television & New Media 14: 440–56. doi: 10.1177/1527476412436 986

Frow, John. 2006. Genre. The New Critical Idiom. London: Routledge.

“Funimation Entertainment—Genres.” 2017. Funimation (blog). Accessed July 25, 2017. https://www.funimation. com/genre/

Green, Rebecca. 2017. “Facet Analysis and Semantic Frames.” Knowledge Organization 44: 397-404.

Halsall, Jane. 2004. “The Anime Revelation.” School Library Journal 50, no. 8: suppl. 6-13.

Hjørland, Birger. 2013. “Facet Analysis: The Logical Ap-proach to Knowledge Organization.” Information Pro-cessing & Management 49: 545-57.

IMDb.com, Inc. Accessed July 24, 2017a. “IMDb.com, Inc. – Browse Popular Genres.” http://www.imdb. com/feature/genre/

IMDb.com, Inc. Accessed July 24, 2017b. “Doraemon (TV Series 1973-).” http://www.imdb.com/title/tt1032831/

IMDb.com, Inc. Accessed July 24, 2017c. “My Neighbor Totoro (1988).” http://www.imdb.com/title/tt00962 83/

Kodansha’s Essential Kanji Dictionary: A Compact Character Dic-tionary for Students and Professionals. 2002. S.v. “青.” Tokyo: Kodansha International.

Kodansha’s Furigana Japanese Dictionary, Japanese-English, English-

Japanese. 1999. S.v. “せいねん青年せいねん

.” Tokyo: Kodansha

International. La Barre, Kathryn. 2010. “Facet Analysis.” Annual Review

of Information Science and Technology 44: 243-84. Lee, Jin Ha, Yuna Shim, and Jacob Jett. 2015. “Analyzing

User Requests for Anime Recommendations.” In JCDL '15 Proceedings of the 15th ACM/IEEE-CS Joint Conference on Digital Libraries Knoxville, Tennessee, USA—June 21=25, 2015. New York: ACM, 269-70 doi>10.1145/ 2756406.2756969.

Levi, Antonia. 2013. “The Sweet Smell of Japan: Anime, Manga, and Japan in North America.” Journal of Asian Pacific Communication 23: 3-18.

MyAnimeList, LLC. Accessed August 1, 2017. “Anime Search.” https://myanimelist.net/anime.php

Poitras, Gilles. 2002. Anime Essentials: Everything a Fan Needs to Know. Berkeley, CA: Stone Bridge Press.

Poitras, Gilles. (2007) 2010. s.v. “Seinen 青年.” Anime and Manga Terminology. Retrieved from http://www.koyagi. com/Terminology.html

Rafferty, Pauline. 2012. “Epistemology, Literary Genre and Knowledge Organization Systems.” In 20 Años del Capítulo Español de ISKO: Actas del X Congreso ISKO Capítulo Español, Ferrol, 30 de junio-1 de julio del 2011, ed. Pérez Pais, María del Carmen, González Bonome and María de los Ángeles. A Coruña: Universidade da Coruña, Servicio de Publicaciones, 553-65.

Ressler, Karen. 2016. “Report: Anime Industry Up 12% in 2015.” Anime News Network (blog), September 29. http://www.animenewsnetwork.com/news/2016-09-



29/report-anime-industry-up-12-percent-in-2015/.107 055 Richmond, Simon. 2009. The Rough Guide to Anime. Rough

Guides Reference. New York: Rough Guides. Robbins, Laura Pope. 2014. “Bringing Anime to Academic

Libraries: A Recommended Core Collection.” Collection Building 33: 46-52.

Sardar, Ziauddin and Borin Van Loon. 2015. Introducing Me-dia Studies: A Graphic Guide. London: Icon Books.

Shiri, Ali. 2014. “Making Sense of Big Data: A Facet Anal-ysis Approach.” In Knowledge Organization 41: 357-68.

Slavic, Aida and Sylvie Davies. 2017. “Facet Analysis in UDC: Questions of Structure, Functionality and Data Formality.” In Knowledge Organization 44: 425-35.

Smiraglia, Richard P. and Joshua A. Henry. 2016. “Facets Among the Topoi: An Emerging Taxonomy of Silent Film Music.” In Knowledge Organization for a Sustainable World: Challenges and Perspectives for Cultural, Scientific, and Technological Sharing in a Connected Society: Proceedings of the Fourteenth International ISKO Conference 27-29 September 2016 Rio de Janeiro, Brazil, ed. J.A. Chaves Guimarães, S. Oliveira Milani, V. Dodebei. Advances in Knowledge Organization 15. Würzburg: Ergon, 153-60.

Spiteri, Louise F. 1998. “The Use of Facet Analysis in In-formation Retrieval Thesauri: An Examination of Se-lected Guidelines for Thesaurus Construction.” Catalog-ing & Classification Quarterly 25: 21-37.

Svenonius, Elaine. 1989. “Design of Controlled Vocabu-laries.” Encyclopedia of Library and Information Science 45 (suppl. 10) , ed. Allen Kent. New York: Marcel Dekker, 82-109.

Svenonius, Elaine. 2000. The Intellectual Foundation of Infor-mation Organization. Digital Libraries and Electronic Publishing. Cambridge, MA: MIT Press.

Thomas, Stefanie. 2014. “Sekai-kei as Existentialist Narra-tive: Positioning Xenosaga within the Genre Frame-work.” MA thesis., The Ohio State University.

Tudor, Andrew. 1974. Theories of Film. Cinema One 23. London: Martin Secker & Warburg.

Turco, Lewis. 1999. The Book of Literary Terms: The Genres of Fiction, Drama, Nonfiction, Literary Criticism, and Schol-arship. Hanover, NH: University Press of New England.

Vickery, B. C. 1960. Faceted Classification: A Guide to Construc-tion and Use of Special Schemes. London: Aslib.

Viz Media, LLC. Accessed July 25, 2017. “Watch.” https:// www.viz.com/watch

Wikimedia Foundation, Inc. Accessed December 17, 2017a. “Doraemon (1979 nen no terebi anime).” ドラえもん (1979 年のテレビアニメ ). https://ja.wikipe-dia.org/wiki/ドラえもん_(1979年のテレビアニメ)

Wikimedia Foundation, Inc. Accessed December 17, 2017b. “Doraemon (2005 nen no terebi anime).” ドラえもん

(2005 年のテレビアニメ ). https://ja.wikipe-dia.org/wiki/ドラえもん_ (2005年のテレビアニメ)

Wikimedia Foundation, Inc. Accessed December 17, 2017c. “Kodomomuke Anime.” 子供向けアニメ . https:// ja.wikipedia.org/wiki/子供向けアニメ#乳幼児・幼

児向けの特徴 Wolfreys, Julian, Ruth Robbins and Kenneth Womack. 2016.

Key Concepts in Literary Theory. [Abington]: Routledge. Appendix: Complete List of Anime Genre Foci Terms Audience Josei (Women) Seinen (Youth) Shōnen (Boys) Shōjo (Girls) Kodomo (Children) Family Setting Setting-Time Cyberpunk Steampunk Historical Setting-Situational Fantasy Apocalypse Science Fiction Science Fantasy Dystopia Setting-Place Restaurants/Stores Isekai (Parallel Universe) Western Space Opera School Afterlife Space Mood Tragedy Thriller Comedy Psychological Horror Suspense Iyashikei (Healing) Characters Magic Idol Singers Gangs Detectives Police



Students Idols Maids Pirates Hunters Hybrids Trap Cyborgs Androids Robots/Mecha Bishōjo (Beautiful Girls) Bishōnen (Beautiful Boys) Monsters Aliens Vampires Busty Females Shotakon (Shōtarō Complex) Rorikon (Lolita Complex) Gender Bender Samurai Demons Delinquents Mahō Shōjo (Magical Girls) Ninja Gods Angels Hero (Protagonist) Heroine (Protagonist) Plot/Narrative Sudden Girlfriend Appearance Slice of Life Coming-of-Age Drama Adventure Mystery Conspiracy Sekai-kei Absurd Abstract Mind-Twisting Gekiga (Plot) Survival Biography Romance Plot/Narrative-Romance-Types Bara (Boys' Love for Male Audience) Shōnen Ai/Yaoi (Boys' Love) Shōjo Ai/Yuri (Girls' Love) Age Gap Plot/Narrative-Romance-Multiple People Love Triangle Harem/Reverse Harem

Subjects Game Vehicles Cooking Time Travel Magic Virtual Reality Tournaments Politics Sexual Contents Violent Contents Work Life Body Sharing/Body Swapping/Reincarnation Crime Music Arts Natural Disaster Amnesia War Military Club Genetic Modification Human Enhancement Superpowers Otaku Culture Abuse Supernatural Seishun (Youth) Terrorism Educational Gambling Mythology Bullying Sports Subjects-Sports Wresting Baseball Gymnastics Boxing Martial Arts Archery Fishing Swimming Volleyball Ice Skating Cycling Soccer Tennis Dancing Association with Other Works Parody Novel



Comics Manga DocumentaryFeature Action Scenes Fan Service Power Suits Henshin (Transformation) Feature-Action Scenes-Types Proxy Battles Swordplay Gunfights Production Production-Techniques CG Animation Production-Styles Chibi (Super Deformed) Gekiga (Style)

Production-Regions Foreign Anime Cartoon Production-Agents Voice Actors Production-Formats ONA (Original Net Animation) Special (SP) OVA (Original Video Animation) Short (Length) Movies Series Production-Languages Dubbed


Sh. Shakti Ghosh, S. Das, and S. Kumar Chatterjee. Human-centric Faceted Approach for Ontology Construction

31

Human-centric Faceted Approach for Ontology Construction†

Shiv Shakti Ghosh*, Subhashis Das**, and Sunil Kumar Chatterjee***

Jadavpur University—Library & Information Science—Main Campus. 188, Raja S.C. Mallick Rd , Kolkata, West Bengal 700032, India,

*<[email protected]>, ***<[email protected]> **University of Trento—ICT-Doctoral School—Via Sommarive, 9 I-38123 Povo, Trento,

TRENTO 38123, Italy, <[email protected]>

Shiv Shakti Ghosh is doing his PhD and is a senior research fellow in the Department of Library & Information Science, Jadavpur University, India. In 2013, he completed his bachelors in mathematics from Serampore College, India and obtained his master’s degree in library and information science (MSLIS) from the Documentation Research and Training Centre (DRTC), Indian Statistical Institute, India in 2015. His research interests include information science and arts and humanities.

Subhashis Das currently works as a post-doctoral researcher at the Department of Information Engineering and Computer Science (DISI), Università degli Studi di Trento, Italy. He obtained his Ph.D from the ICT-Doctoral School, University of Trento, Trento, Italy. Subhashis does research in information science, computing in health science, arts and humanities and geoinformatics (GIS).

Sunil Kumar Chatterjee is a professor in the Department of Library & Information Science, Jadavpur University, India. In 1987, he obtained his MSc in chemistry. In 1990, he got the associateship in information science from INSDOC, India and completed his master’s degree in library and information science from Burdwan University, India in 1991. He obtained his PhD in 2005. Having begun as a researcher in library and information science during the early nineties, he has contributed papers in several peer reviewed journals and conferences of national and international repute. His research areas include scientometrics, informetrics and information science.

Shiv Shakti Ghosh, Subhashis Das, and Sunil Kumar Chatterjee. 2020. “Human-centric Faceted Approach for Ontology Construction.” Knowledge Organization 47(1): 31-44. 39 references. DOI:10.5771/0943-7444-2020-1-31.

Abstract: In this paper, we propose an ontology building method, called human-centric faceted approach for ontology construction (HCFOC). HCFOC uses the human-centric approach, improvised with the idea of selec-tive dissemination of information (SDI), to deal with context. Further, this ontology construction process makes use of facet analysis and an analytico-synthetic classification approach. This novel fusion contributes to the originality of HCFOC and distinguishes it from other existing ontology construction methodologies. Based on HCFOC, an ontology of the tourism domain has been designed using the Protégé-5.5.0 ontology editor. The HCFOC methodology has provided the necessary flexibility, extensibility, robustness and has facilitated the cap-turing of background knowledge. It models the tourism ontology in such a way that it is able to deal with the context of a tourist’s information need with precision. This is evident from the result that more than 90% of the user’s queries were successfully met. The use of domain knowledge and techniques from both library and information science and computer science has helped in the realization of the desired purpose of this ontology construction process. It is envisaged that HCFOC will have implications for ontology developers. The demon-strated tourism ontology can support any tourism information retrieval system.

Received: 8 May 2019; Revised: 17 June 2019, 8 July 2019, 20 August 2019, 2 September 2019, 25 October 2019, 5 November 2019: Ac-cepted: 15 November 2019

Keywords: ontology, tourism, information, domain, concepts, terms

† The link for accessing the tourismontology.owl file can be found at (https://webprotege.stanford.edu/#projects/00bef201-8f71-4f68-b507-b04d4bd9b99d/edit/Classes)


Sh. Shakti Ghosh, S. Das, and S. Kumar Chatterjee. Human-centric Faceted Approach for Ontology Construction 32

1.0 Introduction Recent years have witnessed an increase in the use of ontol-ogy for knowledge representation, sharing and distribution. As defined in Studer et al. (1998), ontology is a formal, ex-plicit specification of a shared conceptualization. Concepts belonging to a domain of discourse are described with the help of properties. The description present in the properties refers to the different features and attributes of the con-cepts. Thus, semantic relationships are established between the concepts. As an outcome, background knowledge or rel-evant semantic information pertaining to the domain of dis-course gets encapsulated. In totality, an ontology attempts to model a domain of discourse. The modeling here refers to either an attempt to describe or categorize objects be-longing to the domain of discourse.

As remarked by Smiraglia (2015, 19), knowledge organ-ization studies are focusing towards a domain-analytical di-rection. Hjørland (2017) has termed domain analysis as the theorization and analytical approach to library and infor-mation science and knowledge organization. For domain analysis, many ontology construction methodologies have conceived the concept of facet analysis (see study by (Prieto-Diaz 2003), (Giunchiglia et al. 2009) and (Das and Roy 2016)) and analytico-synthetic classification pro-pounded by Ranganathan (1967). The analytico-synthetic approach consists of two phases. In the first phase, known as the analysis phase, compound and complex ideas are fragmented into fundamental ideas. After analyzing their characteristics, these fundamental ideas are grouped or clustered according to similarity. This first phase is com-pleted by following the first-link-downwards and last-link-upwards approach. The difference between these two ap-proaches are characterized in the way they start approach-ing concepts, where a concept, as defined by Dahlberg (1978) is a knowledge unit, and the statements about its referent are the characteristics of the given concept. The first-link-downwards starts the analysis from the root con-cepts in the domain and then gradually narrows down to more specific concepts. Thus, it proceeds from abstract level to a concrete level. The last-link-upwards technique identifies and studies the characteristics of base concepts and assembles them depending upon their similarity of features. In this way, by continuing the process of clubbing together a large or universal concept is formed. By repeat-ing this process, the root concept is reached. Thus, the bot-tom up approach proceeds from concrete level to an ab-stract level. In the second phase, mentioned as the synthe-sis phase, semantic relationships are established between concepts. This approach helps in identifying facets, where a facet, as has been described in Giunchiglia and Dutta (2011) as a hierarchy of homogenous group of terms (nodes), each term denoting a primitive atomic concept.

Plenty of research is being conducted to determine con-text. The possibility of using ontology as a tool for context management has also been endeavored. The methodologies for developing or constructing these ontologies depict the human-centric approach to deal with context, its analysis and development. Further, in the field of library and infor-mation science, the idea of selective dissemination of infor-mation (SDI) propounded by Luhn (1961) has been in use for quite a considerable amount of time. SDI has been rou-tinely used to deal with “context.” SDI got involved with the aim of catering information to those who found it most use-ful, or, in other words, it aimed to cater information accord-ing to the right context. Additionally, the aim was to prevent communication of misinformation or, information that is out of the context. The system is to be fed with a database of users’ profiles containing areas of interest belonging to individual end users. SDI advocates that when a particular item or information is to be disseminated, it is to be done after comparing its information pattern or trend with pro-files of the members present in the system. The system may select as many recipients as it finds suitable. There is a con-cept of weeding out imbibed within SDI as it has been rec-ommended to delete those document patterns that have be-come obsolete. Furthermore, interests sustaining for longer time must be preserved. This sense of weeding out of ob-solete information patterns and preservation of infor-mation sustaining for longer time periods has particularly motivated the evaluation step in HCFOC. New concepts and its representatives must be sustained for longer periods and must be inserted and embedded within the ontology with the objective to saturate it. The requirement of addi-tion or deletion of outdated information patterns must be identified by repeating the evaluation step of HCFOC. The idea of SDI is quite similar to the human-centric approach followed while dealing with context. Lamsfus (2009) has fa-miliarized the human-centric approach with relevance to contextual information. The proposed human-centric fac-eted approach for ontology construction (HCFOC), dis-cussed in this paper, uses a synthesis of both these similar approaches to correctly deal with context. Further, the ana-lytico-synthetic approach has been also utilized for design-ing this methodology. Tourism has been chosen as the do-main for exemplifying this process. Since people moving from one place to another may need relevant information for enhancing their mobility at any time, it becomes crucial to accurately determine the context of a tourist’s infor-mation requirement or information seeking pattern. Opting tourism as a domain for study and experimentation is readily providing a scope to test the context-modeling capability of the HCFOC methodology. Smiraglia (2015, 19) has listed the domains that have been taken up recently for knowledge organization studies, which clearly reveals that not even once the tourism domain was taken up for study.



Furthermore, many regions and countries rely on tour-ism as the main source of revenue generation, which con-tributes to the national GDP. This domain has become the focus of a lot of economic activities. According to the World Tourism Organization (UNWTO) (2017), “the business volume of tourism equals or even surpasses that of oil exports, food products or automobiles.” Besides this, it has direct effects on educational, cultural and social sectors. Transportation, hospitality and entertainment ser-vices also harness benefits from this industry. Mobility of people has increased significantly over time. Quite natu-rally, the tourism domain is also experiencing a surge in information and knowledge handling like never before. And to deal with this, innovative approaches and applica-tions are required. So, an effort has been made to build a model for the tourism domain using ontology.

The rest of the paper is organized as follows: Section 2.0 provides a brief literature review on this domain and discusses the related works. Section 3.0 explains the HCFOC methodology exemplifying its use for construct-ing the tourism ontology. Section 4.0 concludes the paper while discussing avenues for future work. 2.0 Ontologies related to tourism Since the efficiency of ontologies in decision making has been proven, a lot of research is focusing on ontology con-struction methodologies. Qiu et al. (2018) have used a combination of rule-based (for concept and relationships extraction), statistics-based (for ranking the concepts) and cluster-based methods (for clustering and constructing taxonomy) for constructing ontologies. Nguyen and Lu (2016) have developed ontologies for web pages. The steps followed are requirement analysis, conceptualization and implementation. Yang et al. (2017) have proposed DOCM or domain ontology construction method. The methodol-ogy involves requirements and domain knowledge analysis followed by establishment, evaluation and modification ef-fort assessment of the ontology. Further, a method has been proposed to evaluate the modification effort on the ontology. Suárez et al. (2015) have developed the NeOn Methodology framework based on glossary of processes and activities and ontology building scenarios, networks and life-cycle models.

The tourism domain is also experiencing a surge in the use of ontologies for information dissemination, decision making and fabrication of recommendation systems. Chu et al. (2016) have constructed a tourism recommender system. For this, users in the database have been categorized as re-lated and unrelated. Different kinds of algorithms have been used to deal with the contexts of users belonging to differ-ent categories. Al-Hassan and Lu (2015) have discussed the use of ontology and the defined relationships and attributes

within it to find semantic similarities between items for use in an e-Government tourism service recommendation sys-tem. The Harmonise ontology proposed by Fodor and Werthner (2005) focuses on tourism data exchange. The QALL-ME framework by Ferrandez et al. (2011) has been mapped with Princeton WordNet by Miller (1995) and the Suggested Upper Merged Ontology (SUMO) and thus it has a strong foundation knowledge base. GETESS by Staab et al. (1999) deploys natural language processing (NLP) and se-mantic web methods to answer user queries using web-based information exchange and distribution. Existing on-tologies on the tourism domain have been listed in Mathur et al. (2015). The Mondeca ontology has been built using concepts from the thesaurus developed by the World Tour-ism Organization (UNWTO) (2001). The OnTour ontology by the eTourism Semantic Web Portal describes the con-cepts of location coordinates, accommodation, date and time of certain events, etc. for tourism information dissem-ination. The purpose of the HiTouch Ontology and the TAGA ontologies is to cater to travel agents. Gregor et al. (2016) have proposed a methodology using semantic clus-tering algorithms to create ontology for intelligent transpor-tation systems. Frikha et al. (2016) and Lee et al. (2017) have shown the use of ontology for medical tourism and leisure tour recommender systems respectively. She et al. (2018) have deployed property graph ontology for a tourism rec-ommender system. Special efforts for understanding and modeling the context in tourism can be seen in (Kashevnik 2017). 3.0 The HCFOC methodology and the tourism

ontology The human-centric faceted ontology construction (HCFOC) methodology consists of eight steps. The on-tology construction process demonstrates the synthesis of the idea of SDI from the field of library and information science with the idea of human-centric approach to deal with the context of the primary information seeker. Fur-ther, the analytico-synthetic classification approach has been also used to capture the necessary and relevant back-ground or inherent knowledge. This methodology evolved while attempts were being made to develop an ontology for the tourism domain, which has been also shown here. The tourism ontology has been developed using the Pro-tégé-5.5.0 (https://protege.stanford.edu) ontology editor. It is a free, open-source ontology editor developed by the Stanford Center for Biomedical Informatics Research at the Stanford University School of Medicine. The Protégé OWL (web ontology language) ontologies consist of clas-ses, properties (object properties and data properties) and individuals. Classes are sets of individuals that are objects of the domain. Object properties are relations between the



objects. Data properties are relations between the objects and data types. 3.1 Step 0: domain selection A field or area of study is to be selected on which the on-tology will be created. The field or area of study is com-monly referred to as domain. The tourism domain is the centre of a lot of economic activities. Contextual infor-mation is highly sought after in this domain. The aim of the HCFOC methodology is to understand and deal with context in a comprehensive and precise manner. Based on previous studies and reviewing of existing literature the tourism domain has persistently appealed as a perfect do-main for testing and implementing this methodology. Moreover, as this domain is multifaceted, the analytico-synthetic approach ingrained in the HCFOC methodology will also be tested. 3.2 Step 1: focus map creation The contextual queries of the primary information seeker in the selected domain are usually considered to ascertain the focus. Ascertainment of the focus helps in aligning with or modifying the purpose and scope of the ontology. This step is instrumental in making the ontology capable of deal-ing with the context of the primary information seeker. Cor-rect and comprehensive understanding of the context helps in increasing the precision of the answers, responses and re-sults obtained in return. To achieve this, the HCFOC meth-odology uses a synthesis of two ideas, namely, the human-centric approach, to deal with context, and selective dissem-ination of information (SDI). Following this synthesis, a prospective map of depicting the contextual behavior of the primary information seeker has been included in this step. This map also represents the purpose and scope of the on-tology under construction.

Competency questions were obtained from prospective tourism information users, which include professors, re-search scholars, students and heads of travel agencies from India and Italy. Some of the frequently asked questions are: a) Where can I stay during my visit to Kolkata?; b) Which rivers pass through India?; c) List all the mountains in India?; d) List of cuisines of India?; e) List of bridges in India?; f) Which deserts are located in India?; g) How to reach Mum-bai from Kolkata?; h) What is the local language of Salem, Tamil Nadu?; i) How is the weather of Bangalore in June?; and, j) What are the drinks available in Goa?

Accordingly, to answer the aforementioned questions, some of the terms that need to be considered are: address, administrative division, drinks, alcohol, artifact, structure, bridge, bus, car, transport, city, contact, email, fax, country code, country, geo-coordinate, location, height, hotel, lan-

guage, weather, latitude, longitude, mountain, landform, desert, postal code, river, basin, vehicle, website, etc. An analysis of these questions revealed the approach and in-formation seeking behavior in this domain, and the pri-mary information seeker was identified as a tourist. So, this ontology has been conceived keeping the tourist at the centre. Further, this analysis also helped us in forming a prospective map of the context of tourists’ information needs. This map also represents the purpose and scope of the tourism ontology. 3.3 Step 2: information acquisition A footprint of the ideas, concepts and their features, pre-sent in the existing information sources on the domain un-der consideration, is to be formed. This footprint is to be matched against the map created in the previous step, and the overlapping areas are to be identified. Information sources on all such overlapping areas are to be studied and the collected information must be consolidated.

Figure 1(a) depicts the general information acquisition process, and Figure 1(b) shows how tourism information is scattered over different domains like geography, cadas-tre, automotive and economy. Our present requirements are only a subset of information from each domain. For instance, monuments and heritage buildings are usually collected from the cadastre database, but we might not need property tax or building material information in the context of tourism application.

Information for developing the tourism ontology has been obtained from different governmental and non-gov-ernmental tourism websites, reference tools on tourism, travelers and users of tourism information. Several indi-vidual travel experiences that were shared over the internet were studied. Information provided by users in travel groups prevalent on social media platforms were also scru-tinized. The results of the assessment done on these re-sources have been explained in Ghosh and Chatterjee (2019). A reliable information base was formed after com-piling all such information. 3.4 Step 3: term identification Terms representing the ideas, concepts and their features belonging to the overlapping areas found in the previous step are to be identified. Different tourism related terms were identified from the compiled information base formed in the “information acquisition” step. The World Tourism Organization (UNWTO) thesaurus (World Tour-ism Organization 2001) has been used as a guide to tour-ism terminology. WordNet was rigorously consulted in this process. Besides this, Schema.org (https://schema.org/ Thing) and the INSPIRE (https://inspire.ec.europa.eu/)



knowledge base were also consulted. This helped in resolv-ing the ambiguity involved in understanding the concepts. Also, the perfect term to represent the concepts could be identified. Some of the terms are: artifact, structure, lodg-ing, hostel, monument, telpherage, biome, cuisine, drink, event, mountain, hill, plateau, location and person. 3.5 Step 4: analysis. The terms obtained in the previous step have to be ana-lyzed for identification of differences and similarities. Terms denoting compound and complex ideas have to be disintegrated into terms denoting simple ideas. Keeping in mind the purpose, scope and context of the primary infor-mation seeker, it is to be analyzed whether a term will be used to denote a class / sub-class / instance / relation or attribute (of object or of data).

For the tourism ontology, the analysis was continued with reckon to the purpose, scope and context of a tourist. Besides other principles as proposed in Ranganathan (1967), the “principle of context” and “canon of rele-vance” was widely used in the procedure. Terms with sim-ilar features were grouped together. For example, terms like arena, camp, hospice and hostel were found to be sim-ilar. They have been grouped together.

Analysis has been done without user participation as it difficult to involve users. This is because not all users are domain experts. Many of them are leisure travelers. Some of them are also not aware of the exact piece of infor-mation they want (very similar to users in libraries, where often the librarians try to find out the exact needs of the users). Certain categories of users have been involved for defining the competency questions and their respective evaluation. Many user centric questions were formulated, which proved to be helpful in designing and evaluating the

tourism ontology. Domain expertise and proficiency and friendliness with the vocabularies and tools used in the study were an essential requirement of this step. Due to lack of required ability, not many users were enthusiastic or eligible to participate in this step. Some were also wary of the amount of time this work would require. Thus, though we had initially thought of involving the users in this step, we dropped this idea later. 3.6 Step 5: knowledge synthesis After the analysis carried out in the previous step, the ideas were clustered or grouped together based on similarity in characteristic, and the categories have to be labeled. In this step, for classifying the facets, the first link downwards and last link upwards approach was followed. The first link downwards approach proceeds from abstractness to con-creteness. While the last link upwards approach proceeds from specific concepts towards generic concepts. Following these two approaches leads to an overall increase in the de-gree of robustness. Knowledge on the domain is synthe-sized in this step by establishing relationships between the concepts.

For the tourism ontology, facet discovery and inventory control were guided by the “principle of context” and “prin-ciple of helpful sequence” as proposed in Ranganathan (1967). For example, the group containing the terms arena, camp, hospice and hostel were listed under “lodging.” The features of the concepts represented in one group were used to establish relationships between concepts repre-sented in other groups. For example, “lodging” has features like price, location etc., which have been used to establish its relationship with other concepts. Facets contain sub-facets. For example, “lodging” has been listed under “structure,” which has been again listed under “artifact.”

Figure 1(a). Information acquisition Figure 1(b). Domains intersecting with tourism



3.7 Step 6: knowledge representation This step consists of the following sub steps: Term standardization: There may exist many terms to

denote a single concept. However, such synonymous terms may differ among themselves based on usage. The use of an appropriate and proximate term has to be decided. Such proximate terminology should suit the context of the ontology model. Also, the terms chosen should be most frequently used or collected from a standard vocabulary on the domain on which the ontol-ogy is being constructed. Use of popular terms in-creases user friendliness while use of terms from stand-ard vocabulary helps in interoperability. Use of a term by domain experts in their written and verbal commu-nication influences its selection for use in the ontology.

The terms that have been standardized for use in the tourism ontology have been enlisted by consulting dif-ferent controlled vocabularies on tourism. For example, the word “artifact” has been chosen to represent the concept “any object made by human beings, especially with a view to subsequent use.” The term “artifact” has been chosen from those terms representing the sense of product such as, merchandise, produce, creation; ex-amples of terms representing the sense of relic such as, antique, monument (representing the sense relic); ex-amples of terms representing the sense such as, handi-work, artisanship (representing the sense handicraft).

Ordering: There exist many criteria for ordering the standardized terms within the array. Some of the crite-ria as mentioned in Ranganathan (1967) are existent classification schemes on the domain, alphabetical or-der, decreasing or increasing complexity, increasing or decreasing extension, etc. Whichever criterion is cho-sen, it must be kept in mind that the ordering must help in reaching the desired aim of the ontology.

This step was completed keeping in mind the purpose and scope of the tourism ontology and the context of the tourist’s information need. In Table 1, the outcome of ordering is visible.

Modeling: The HCFOC methodology uses the DERA (domain, entity, relation and attribute) framework men-tioned in Giunchiglia and Dutta (2011) for structuring the facets of the domain under consideration. Here, the idea of entity has been derived from Bhattacharyya (1975). Ranganathan’s faceted classification (1989) di-vides knowledge into five fundamental categories, namely, “personality” (P), “matter” (M), “energy” (E), “space” (S) and “time” (T), known by the acronym PMEST. Bhattacharyya (1981) further refined the divi-sion into four main categories, namely, “discipline” (or domain) (D), “entity” (E), “property” (P) and “action”

Class Thing Event MentalObject Cuisine AfricanCuisine … PhysicalObject Artifact Handicraft Structure Building HealthcareFacility Hospital … Brothel Library … Fountain … TransportationSystem AirTransportationSystem … Location AdministrativeDivision … Person TravelAgent … Substance Drink …

Object Properties hasCreator isBasedFrom hasArrivalPoint …

Data Properties hasPrice hasIdentifier hasGeocoordinate … hasCheckInTime hasRating …

Table 1. Partial list of classes, object properties and data properties from the tourism ontology.

(A), and an additional special category called “modifier”

(m), known by the acronym DEPA. The DERA frame-work advocates organization of knowledge into do-mains. Further, each domain should be organized using facets. Giunchiglia et al., (2014) shows that DERA allows addition of domains, facets and terms into the ontology, whenever required, and its exploration for automatic rea soning via direct encoding into description logics (DL).



Due to these features of DERA, the characteristic fea-tures of DL, like, soundness, decidability and decision procedures will be inherited into the system. Further-more, since addition of facets can be done any time, the use of DERA makes the system expandable.

All that has been developed in the previous steps are to be put into action in this penultimate step using DERA. The domain knowledge that has been synthesized in the previous steps is to be expressed in this step by clearly establishing the relationships between the concepts. For this ontology on the tourism domain D, the set of fac-ets Event, MentalObject, PhysicalObject, Substance, etc. belong to the element E, the set of facets hasCrea-tor, hasIdentifier, isBasedFrom, isLocatedAt, etc. be-long to the element R and the set of facets Cost, Name, Currency etc. belong to the element A. An example of an established relationship is, Taj Mahal (Subject) isLo-catedAt (Predicate) Agra (Object). Similarly, Feni (Sub-ject) isBasedFrom (Predicate) Goa (Object). Here, Taj_Mahal, Agra, Feni, Goa are instances of the classes Monument, AdministrativeDivision, AlcoholicDrink and AdministrativeDivision respectively.

3.8 Step 7: evaluation In this step, the ontology is put to test. It is to be tested whether the ontology is fulfilling its purpose according to the specifications. It is to be found out whether the ontol-ogy is able to deal with the context intended for, thereby addressing other aims and objectives. Necessary correc-tions are to be made, if required, by going back and re-peating the steps in order.

The syntactic correctness and consistency of the tour-ism ontology were checked in Protégé, using the HermiT OWL reasoner. The HCFOC methodology itself ensures the completeness and conciseness of the tourism ontol-ogy. The usability of the tourism ontology from a tourist’s point of view has been gauged. It has been found out whether the ontology is capable of understanding the tourist’s context with the help of competency queries. Use of competency queries as an evaluation method is one of the best available methods to evaluate an ontology, as has been suggested by Abacha et al. (2013) and Bezerra et al. (2013). Competency queries provided the way to check the entity (E) facet, relation (R) facet and attribute (A) facet together, which are embedded in the form of natural lan-

Figure 2. Flowchart of HCFOC methodology.



guage in a given question. For example, queries like “What are the amenities (X) provided by the hotels (Y) nearby to the place?” From this natural language question, we can derive “identification” using a general query pattern: Give me all X in Y AND WHERE.property.True. Iden-

tification: “concepts” and “properties.” Entity: hotel, place relation (R) nearby, place and attribute (A) amen-ities. Boolean.

The evaluation of the tourism ontology was carried out by research scholars and students belonging to Jadavpur Uni-versity, Kolkata who are quite enthusiastic when it comes to tourism. Queries collected from them helped in check-ing the elements entity, relation and attribute embedded in natural language together. They have analyzed whether the ontology is able to meet their criteria of needs. The group of evaluators and the group from whom the competency questions had been obtained were disjoint. The evaluators were asked to imagine that they are going to visit a place as a tourist. Now, based on the above situation, they were asked to enlist the questions whose answers they would like to know before the visit.

The purpose of the ontology constructors behind set-ting up this situation and asking the evaluators to complete

the task was to list as many user queries as possible. This helped in understanding the appropriateness of the ontol-ogy framework. This also helped in finding out whether the ontology could meet the purpose for which it was built. The comments posted by the evaluators served as a poten-tial feedback on the usefulness of the tourism ontology and how it could be improved further. The listed questions provided an in-depth insight on the information seeking behavior of the tourists.

Key terms were extracted from the questions listed by the users. Then, it was found out manually for each ex-tracted key term whether, the term, or a synonymous term, or the concept denoted by the term is present in the on-tology or not. The questions with all key terms having rep-resentation in the ontology were marked as fully answera-ble. The questions with some key terms having represen-tation in the ontology were marked as partially answerable. The questions with all key terms having no representation in the ontology were marked as not answerable. The fully answerable, partially answerable and not answerable ques-tions were pointed out to the evaluators after the marking. 90.19% of the queries posed by the evaluators were fully answerable. 2.94% of the queries were partially answera-ble. 6.86% were unanswerable. The concepts denoted by the key terms present in the questions that were partially

Query Extracted key terms or concepts

What is the currency of the place? <currency, location>

What are the festivals that will be held in Paris during the time of visit? <festival, location, time>

Which mode of transport is to be availed to visit the place? <transport, location>

Where to stay during the visit to the place? <accommodation, location>

What are the local foods available at that place? <cuisine, location>

Table 2. List of some of the fully answerable queries and the extracted key terms or concepts.


Which are the tourist spots adjacent to the place? <tourist spot, adjacent, location> Is the place secure for tourists? <location, security, tourists> Are there any nearby markets to the place? <markets, nearby, location>

Table 3. List of some of the partially answerable queries and the extracted key terms or concepts.


Will water be available on the highways connecting two places? <water, highways, location>

What are the amenities provided by the hotels nearby to the place? <amenities, hotels, nearby, location>

What is the socio-political scenario of the place? <socio-political, location>

What are the fields that an educational institution located at a place specializes in? <education, institute, specialization>

How is the mobile network connectivity at the place? <mobile, network, connectivity, location>

Table 4. List of some of the unanswerable queries and the extracted key terms or concepts.



answerable or not answerable were analyzed. Out of the unanswerable questions many were found to be out of the scope of the ontology. For example, “What is the socio-political scenario of the place?” or “What are the fields that an educational institution located at a place specializes in?” Queries expressed using spatial relations like “nearby,” “adjacent,” etc., were either unanswerable or were partially answerable. During the initial conception of the tourism ontology, only geo-coordinates were included for locating a place. But, since users tend to express queries using the natural language sense of distance, hence it was decided that the spatial relations must be included. Some of the spatial relations, as has been mentioned in Dutta et al. (2011), that need to be considered are: directional (north, south, north-east, south-west, etc.), internal (inside, cen-tral, etc.), external (adjacent, nearby, etc.), position with re-spect to a border (overlap, opposite, etc.), longitudinal (be-hind, towards, etc.), sideways (left, right, etc.) and relative (up, below, etc.).

The class visualization of the entities in the tourism on-tology has been done using the ProtégéVOWL (http:// vowl.visualdataweb.org/protegevowl.html) visualization tool. Figure 3 partially shows the hierarchy of the tourism ontology and the class visualization on the left and right side of the figure respectively. The connected entities and the visualization were shown to the evaluators. The key terms from the evaluators’ queries that were visible in the visualization were pointed out to the evaluators.

For query visualization and analytics, the tourism model was deployed using GraphDB (http://graphdb.onto-

text.com) by OntoText, an enterprise-ready semantic graph database, compliant with W3C standards. Figure 4 depicts how the food named “panipuri,” an instance of AsianCuisine is related with the country India by the rela-tion isBasedFrom. On the right side of Figure 4, descrip-tion, type and rank of this namedIndividual are available. Similarly, Figure 5 shows that “pizza,” an instance of Eu-ropeanCuisine isBasedFrom the country Italy.

Then they were asked to comment expressing their con-cerns and suggestions. The reports given by the evaluators once again acted as a tool to determine the usefulness of the tourism ontology and how it could be improved further. This strategy of evaluation ingrained in the HCFOC meth-odology aims to increase the exhaustiveness of the ontol-ogy. After taking into consideration the comments of the evaluators and with a view of expanding the scope and coverage of the ontology many concepts have been noted down. As of now, the tourism ontological model does not contain background knowledge on transport booking sys-tem, price or tax havens (for business tourism where people may seek information for taking advantage of lower prices (for example, booze cruise) or people trying to take ad-vantage of tax loopholes), social welfare systems (for ben-efit tourism where people move to take advantage of wel-fare schemes), law (for people moving to take advantage of the legal system for filing lawsuits), birthright (for tourists with the purpose of giving birth in the destination), hotel amenities (for example, number of suites, air-conditioning), accessibility of disables (website accessibility, trained staff availability for dealing with accessibility issues, well-adapted

Evaluators No. of queries No. of Fully answerable queries No. of Partially answerable queries No. of Not answerable queries

Evaluator 1 9 8 1 0

Evaluator 2 6 6 0 0

Evaluator 3 7 6 0 1

Evaluator 4 7 6 0 1

Evaluator 5 6 5 1 0

Evaluator 6 6 5 0 1

Evaluator 7 7 7 0 0

Evaluator 8 11 10 0 1

Evaluator 9 8 8 0 0







Total 102 92 3 7

Table 5. Statistics of queries posed by the evaluators.



hotel rooms, technical aids and disability equipment such as wheelchairs, bath chairs and toilet raisers, accessible res-taurants and bars, adapted toilets in restaurants and public places, accessible streets and sidewalks, a specific attrac-tion’s level of accessibility), agri-tourism events (for people willing to participate in cattle drives or ranches), adventure sports (for example, rambling, climbing, biking, horseback riding, caving, hiking, trekking, snowshoeing, ski moun-taineering, diving, rafting), drug tourism, fashion tourism, genealogy tourism, halal tourism (pork and alcohol free flights and hotels, separate spa and swimming pools for men and women, announcement of prayer timings and re-

ligious programs), kosher tourism (for orthodox Jews re- quiring kosher foods, accommodations within walking dis-tance from synagogues, flights with kosher meals), literary tourism (dealing with places and events from fictional texts and their authors’ lives, for example, Tolkien tourism by fans of The Lord of the Rings), romance tourism (for people travelling in search of relationship), sex tourism (for peo-ple travelling to have sex), set-jetting (for people traveling to destinations first seen in movies), medical tourism or wellness tourism (for people travelling to obtain medical treatment or improve health focusing on prevention), sui-cide or euthanasia tourism (for people traveling to commit

Figure 3. Connected entities.

Figure 4. Visualization of individuals panipuri and India.



suicide or assisted suicide), 3DVT (or 3D virtual tourism, for people willing to explore physical places without phys-ical travel) or tombstone tourism (for people travelling to visit cemeteries, epitaphs, etc.).

The evaluation strategy also presented some questions that will catalogue the guidelines for any tourism infor-mation system that may be built on top of this ontology. For example, “How much money is required for this tour?” or “Which mode of transport will be cheaper?” To answer such questions, the information system behind which the tourism ontology will be instrumental, must have capabilities of drawing inference. It would be advan-tageous to make use of the resources made available by the LOD (linked open data) project rather than populating the ontology with individuals that would be time-consuming. Such an effort has been seen in Dastgheib, Mesbah and Kochut (2013), where the mOntage framework has been introduced, which allows populating the ontology from se-lected LOD sources. Prototype system architecture has been provided below:

We envisage that the demonstrated tourism ontology constructed following the HCFOC methodology will be able to support any tourism information retrieval system. A prototype system architecture has been provided in Figure 6. In our future work, we intend to implement a prototype system using GraphDB by OntoText (http://graphdb.on-totext.com). It is an enterprise-ready semantic graph data-base, compliant with W3C standards. Semantic graph data-bases (also called RDF triple stores) provide the core infrastructure for solutions where modelling agility, data integra-

tion, relationship exploration and cross-enterprise data pub-lishing and consumption are important. The “connected” graph is the final implementation of the model in the GraphDB platform. Figure 4 and 5 depicts snap shots of the connected graph of HCFOC ontology. From Figure 4 and 5 we can easily understand how one individual is con-nected with other related entities. Where the same color nodes represent entities, which belong to the same class, and directed arrows depict how they are connected. GraphDB also supports queries based on simple structured query lan-guage (SQL) as well as semantic similarity. 4.0 Conclusion and future work The HCFOC methodology demonstrated here depicts the detailed modus operandi followed for building the ontol-ogy. The efficacy of the human-centric context modeling and the faceted approach ingrained within the HCFOC methodology were explicitly visible while building the tourism ontology. The tourism ontology has been built considering the tourist as the primary information seeker in this domain and thus the tourist was at the centre of conception of the ontology. 90% of the queries could be answered just after the initial phase of the ontology con-struction. It is contemplated that after inserting the con-cepts succeeding the initial evaluation phase, the precision of answers returned will further increase. Owing to the ad-vent of the faceted approach, it will be extremely effortless and straight forward to insert concepts into the ontology, in the near future. For example, tombstone tourism can be

Figure 5. Visualization of individuals pizza and Italy.



inserted while considering tombstone under structure and listing down its attributes. Similarly, agri-tourism events, adventure sports can be listed under events. Thus, faceted approach ingrained in the HCFOC methodology accounts for the scalability of the tourism ontology. The HCFOC methodology is non-domain specific and future work in-volves using it to build ontologies for other domains. The tourism ontology is to be expanded and it is also to be found out whether parts of the tourism ontology can be reused. We intend to take up the work of demonstrating the advantages of HCFOC in respect to the other existing domain ontology models or framework. A detailed com-parative and comprehensive study between HCFOC and the ontology construction methodologies covered in our literature review is being planned at a very rudimentary level. In that future work, we wish to include any new on-tology construction methodology that emerges, followed with a discussion on how other usual ontology develop-ment methodologies can improve by taking insights from HCFOC.

References Abacha, Asma Ben, Marcos Da Silveira and Cédric Pruski.

2013. “Medical Ontology Validation through Question Answering.” In Artificial Intelligence in Medicine: 14th Con-ference on Artificial Intelligence in Medicine, AIME 2013, Murcia, Spain, May 29 - June 1, 2013. Proceedings, ed. Niels Peek, Mari ́n Morales, Luis Roque and More Peleg. Lec-ture Notes in Computer Science 7885. Berlin: Springer, 196-205. doi:10.1007/978-3-642-38326-7_30

Al-Hassan, Malak, Haiyan Lu and Jie Lu. 2015. “A Seman-tic Enhanced Hybrid Recommendation Approach: A Case Study of e-Government Tourism Service Recom-mendation System.” Decision Support System 72: 97-109. doi:10.1016/j.dss.2015.02.001

Bezerra, Camila, Fred Freitas and Filipe Santana. 2013. “Evaluating Ontologies with Competency Questions.” In Proceedings of the 2013 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT), Atlanta, GA, USA, Nov 17-20, 2013.

Figure 6. Prototype system architecture.



Washington, DC: IEEE Computer Society, 284-5. doi: 10.1109/wi-iat.2013.199

Bhattacharyya, Ganesh. 1975. “Fundamentals of Subject Indexing Languages.” In Ordering Systems for Global Infor-mation Networks: Proceedings of the Third International Study Conference on Classification Research, Bombay, India. January 6-11, 1975, ed. A. Neelameghan. Bangalore: Sarada Ranga-nathan Endowment for Library Science, 83-9.

Bhattacharyya, G. 1981. “Subject Indexing Language: Its Theory and Practice.” In Proceedings of the DRTC Refresher Seminar-13: New Developments in LIS in India, October 14-17, 1981. Bangalore: Indian Statistical Institute. PAGE

Chu, Yan, Hongbin Wang, Liying Zheng, Zhengkui Wang and Kian-Lee Tan. 2016. “TRSO: A Tourism Recom-mender System Based on Ontology.” In Knowledge Science, Engineering and Management: 9th International Conference, KSEM 2016, Passau, Germany, October 5-7, 2016, Proceed-ings. ed. Franz Lehner and Nora Fteimi. Lecture Notes in Computer Science 9983. Cham: Springer, 567-79. doi:10. 1007/978-3-319-47650-6_45

Dahlberg, Ingetraut. 1978. “A Referent-Oriented, Analytical Concept Theory for INTERCONCEPT.” Knowledge Or-ganization 5: 142-51. doi: 10.5771/0943-7444-1978-3-142

Das, Subhashis and Sayon Roy. 2016. “Faceted Ontological Model for Brain Tumour Study.” Knowledge Organization 43: 3-12. doi:10.5771/0943-7444-2016-1-3

Dastgheib, Shima, Arsham Mesbah and Krys Kochut. 2013. “mOntage: Building Domain Ontologies from Linked Open Data.” In ICSC 2013: 2013 IEEE Seventh International Conference on Semantic Computing: Proceedings, 16-18 September, 2013, Irvine, California, USA, ed. Randall Bilof. Los Alamitos, CA: IEEE Computer Society, 70-77. doi:10.1109/icsc.2013.21

Dutta, Biswanath, Fausto Giunchiglia and Vincenzo Mal-tese. 2011. “A Facet-Based Methodology for Geo-Spatial Modeling.” In GeoSpatial Semantics: 4th International Confer-ence, GeoS 2011, Brest, France, May 12-13, 2011; Proceedings, ed. Christophe Claramunt, Sergei Levashkin and Michela Bertolotto. Lecture Notes in Computer Science 6631. [Berlin]: Springer, Berlin, Heidelberg, 133-50. doi:10.10 07/978-3-642-20630-6_9

Ferrandez, Oscar, Christian Spurk, Milen Kouylekov, Iustin Dornescu, Sergio Ferrandez, Matteo Negri, Ruben Izquierdo, et al. 2011. “The QALL-ME Framework: A Specifiable-Domain Multilingual Question Answering Archtecture.” Web Semantics: Science, Services and Agents on the World Wide Web 9: 137-145. doi:10.1016/j.web-sem.2011.01.002

Fodor, Oliver and Hannes Werthner. 2005. “Harmonise: A Step Toward an Interoperable E-Tourism Marketplace.” International Journal of Electronic Commerce 9: 11-39. doi:10. 1080/10864415.2005.11044324

Frikha, Mohamed, Mohamed Mhiri, Mounir Zarai and Faiez Gargouri. 2016. “Using TMT Ontology in Trust Based Medical Tourism Recommender System.” In IEEE/ ACS 13th International Conference of Computer Systems and Applications (AICCSA), Agadir, Morocco, Nov 29 – Dec 02, 2016. Piscataway, NJ: IEEE. 1-8. doi:10.1109/aiccsa. 2016.7945768

Ghosh, Shiv Shakti and Sunil Kumar Chatterjee. 2019. “Assessment of Digital Reference Tools/Websites Re-lated to Tourism Information Using Webometrics and Qualitative Analysis.” SRELS Journal of Information Man-agement 54: 26-31. doi:10.17821/srels/2019/v56i1/137 324

Giunchiglia, Fausto and Biswanath Dutta. 2011. “DERA: A Faceted Knowledge Organization Framework.” http:// eprints.biblio.unitn.it/2104/

Giunchiglia, Fausto, Biswanath Dutta and Vincenzo Mal-tese. 2009. “Faceted Lightweight Ontologies.” In Con-ceptual Modeling: Foundations and Applications, ed. Alexan-der T. Borgida, Vinay K. Chaudhri, Paolo Giorgini and Eric S. Yu. Lecture Notes in Computer Science 5600. [Berlin]: Springer, Berlin, Heidelberg, 36-51. doi:10. 1007/978-3-642-02463-4_3

Giunchiglia, Fausto, Biswanath Dutta and Vincenzo Mal-tese. 2014. “From Knowledge Organization to Know- ledge Representation.” Knowledge Organization 41: 44-56. doi:10.5771/0943-7444-2014-1-44

Gregor, D., S. Toral, T. Ariza, F. Barrero, R. Gregor, J. Rodas and M. Arzamendia. 2016. “A Methodology for Structured Ontology Construction Applied to Intelli-gent Transportation Systems.” Computer Standards & In-terfaces 47: 108-19. doi:10.1016/j.csi.2015.10.002

Hjørland, Birger. 2017. “Domain Analysis.” Knowledge Or-ganization 44: 436-64.

Kashevnik, A. M., A. V. Ponomarev and A. V. Smirnov. 2017. “A Multimodel Context-Aware Tourism Recom-mendation Service: Approach and Architecture.” Journal of Computer and Systems Sciences International 56: 245-58. doi:10.1134/s1064230717020125

Lamsfus, Carlos, Aurkene Alzua-Sorzabal, David Martín, Zigor Salvador and Alex Usandizaga. 2009. “Human-Centric Ontology-Based Context Modelling in Tour-ism.” In International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (KEOD), Madeira, Portugal, October 6-8, 2009, ed. Jan L. G. Dietz. Setúbal: SciTePress, 424-34. doi:10.5220/00023007042 40434

Lee, Chin-I, Tse-Chih Hsia, Hsiang-Chih Hsu and Jing-Ya Lin. 2017. “Ontology-Based Tourism Recommenda-tion System.” In 4th International Conference on Industrial Engineering and Applications (ICIEA), Nagoya, Japan, April 27-29, 2017. ed. Felix T.S. Chan. Piscataway, NJ: IEEE. 376- 9. doi:10.1109/iea.2017.7939242



Luhn, H. P. 1961. “Selective Dissemination of New Scien-tific Information with the Aid of Electronic Processing Equipment.” American Documentation 12: 131-8. doi:10. 1002/asi.5090120209

Mathur, Archana, Akshatha K, Apoorva Shastry and An-itha J. 2015. “A Survey on Existing Tourism Ontolo-gies.” International Journal of Research in Engineering and Technology 4: 20-3. doi:10.15623/ijret.2015.0426005

Miller, George A. 1995. "WordNet: A Lexical Database for English." Communications of the ACM 38, no. 11: 39-41

Nguyen, Thi Thanh Sang and Haiyan Lu. 2016. “Domain Ontology Construction Using Web Usage Data.” In AI 2016: Advances in Artificial Intelligence: 29th Australasian Joint Conference, Hobart, TAS, Australia, December 5-8, 2016, Pro-ceedings, ed. Byeong Ho Kang and Quan Bai. Lecture Notes in Computer Science 9992. Cham: Springer, 338-44. doi:10.1007/978-3-319-50127-7_29

Prieto-Díaz, Rubén. 2003. “A Faceted Approach to Build-ing Ontologies.” In Proceedings Fifth IEEE Workshop on Mobile Computing Systems and Applications, Monterey, Cali-fornia, October 9-10, 2003. ed. DeeberAzada. Los Alami-tos, CA: IEEE.458-65. doi:10.1109/iri.2003.1251451

Qiu, Jing, Lin Qi, Jianliang Wang and Guanghua Zhang. 2017. “A Hybrid-Based Method for Chinese Domain Lightweight Ontology Construction.” International Jour-nal of Machine Learning and Cybernetics 9: 1519-31. doi:10. 1007/s13042-017-0661-0

Ranganathan, S. R. 1989. Colon Classification. 7th ed., rev. and ed. M.A. Gopinath. Bangalore: Sarada Ranganathan En-dowment for Library Science.

Ranganathan, Shiyali Ramamrita. 1937. Prolegomena to Li-brary Classification. Madras: Madras Library Association.

Schema.org. 2019. “Thing.” Accessed October 24. https:// schema.org/Thing

She, Keng-Seong, Su-Cheng Haw, Yu-Gene Loh and Fang-Fang Chua. 2018. "AK Tourism: A Property Graph On-tology-based Tourism Recommender System." In Knowledge Management International Conference (KMICe), July 25 -27 2018, Miri Sarawak, Malaysia. ed. Fauziah Baharom

Assoc, Yuhanis Yusof, Mazida Ahmad and Rohaida Romli. Changlun, Malaysia: UUM College of Arts and Sciences, 83-8. http://repo.uum.edu.my/25195/

Smiraglia, Richard P. 2015. Domain Analysis for Knowledge Or-ganization: Tools for Ontology Extraction. Waltham, MA: Chandos.

Staab, Steffen, Christian Braun, Ilvio Bruder, Antje Düs-terhöft, Andreas Heuer, Meike Klettke, Günter. Neu-mann, et al. 1999. “GETESS: Searching the Web Ex-ploiting German Texts.” In Cooperative Information Agents III: Third International Workshop, CIA’99 Uppsala, Sweden, July 31–August 2, 1999 Proceedings, ed. Matthias Klusch, Onn M. Shehory and Gerhard Weiss. Lecture Notes in Computer Science 1652. [Berlin]: Springer, Berlin, Hei-delberg, 113-24. doi:10.1007/3-540-48414-0_7

Studer, Rudi, V., Richard Benjamins and Dieter Fensel. 1998. “Knowledge Engineering: Principles and Methods.” Data & Knowledge Engineering 25: 161-97. doi:10.1016/ s0169-023x(97)00056-6

Suárez-Figueroa, Mari Carmen, Asunción Gómez-Pérez and Mariano Fernández-López. 2015. “The NeOn Methodology Framework: A Scenario-Based Method-ology for Ontology Development.” Applied Ontology 10: 107-45. doi:10.3233/ao-150145.\

UNWTO (World Tourism Organization). 2001. Thesaurus on Tourism & Leisure Activities (Trilingual: English, French, Span-ish). [Madrid]: World Tourism Organization. https:// www.e-unwto.org/doi/pdf/10.18111/9789284404551

UNWTO (World Tourism Organization). 2019. “Why Tourism? Accessed October 25. https://www.unwto. org/why-tourism

Yang, Yuehua, Yuan Ping, Junping Du and Hui Ma. 2017. “A Domain Ontology Construction Method with Ontol-ogy Modification Effort Assessment.” In Proceedings of 2017 Chinese Intelligent Automation Conference, Tianjin, China, June 2-4, ed. Zhidong Deng. Lecture Notes in Electrical Engineering 458. Singapore: Springer, 473-80. doi:10.10 07/978-981-10-6445-6_52


A. MacFarlane, S. Missaoui and S. Frankowska-Takhari. On Machine Learning and Knowledge Organization …

45

On Machine Learning and Knowledge Organization in Multimedia Information Retrieval†

Andrew MacFarlane*, Sondess Missaoui**, and Sylwia Frankowska-Takhari***

Centre for HCI Design, Department of Computer Science, City, University of London, UK *<[email protected]>, **<[email protected]>, ***<[email protected]>

Andrew MacFarlane is a reader in information retrieval in the Centre for HCI Design, Department of Computer Science at City, University of London. He got his PhD in information science from the same institution. His research interests currently focus on a number of areas including image retrieval, disabilities and information retrieval (dyslexia in particular) and AI techniques for information retrieval and filtering. He was the principle investigator for the PhotoBrief project, which focused on meta-data and images and is current involved in the DMNIR project, which is investigating information verification tools for journalists.

Sondess Missaoui is a postdoctoral research fellow in information retrieval in the Centre for Human-Computer Interaction Design, City, University of London. She graduated in computer science at the University of IHEC Carthage (Tunisia) and she obtained her PhD in computer sciences at the University of Milano-Bicocca, Depart-ment of Informatics, Systems, and Communication (Italy). Her research interests are recommender systems, information retrieval, mobile information retrieval, context-awareness, and user profiling. Currently, she is work-ing on a research project (DMINR) that aims to create a digital tool for researching and verifying stories. She focuses on a number of areas including aggregated search, natural language processing, and deep learning.

Sylwia Frankowska-Takhari holds a MA in linguistics and information science from the University of Poznan, Poland (2001) and a MSc in human-centred systems from City, University of London (2011). She completed her PhD under the supervision of Dr. Andrew MacFarlane and Dr. Simone Stumpf at the Centre for HCI Design, Department of Computer Science at City, University of London. Her key research interests are information behaviour and image retrieval. Sylwia’s PhD work investigates the information behaviour and image needs of professionals working in creative industries with a particular focus on how images are selected, used and tailored in online journalism.

MacFarlane, Andrew, Sondess Missaoui and Sylwia Frankowska-Takhari. 2020. “On Machine Learning and Knowledge Organization in Multimedia Information Retrieval.” Knowledge Organization 47(1): 45-55. 44 refer-ences. DOI:10.5771/0943-7444-2020-1-45.

Abstract: Recent technological developments have increased the use of machine learning to solve many prob-lems, including many in information retrieval. Multimedia information retrieval as a problem represents a signif-icant challenge to machine learning as a technological solution, but some problems can still be addressed by using appropriate AI techniques. We review the technological developments and provide a perspective on the use of machine learning in conjunction with knowledge organization to address multimedia IR needs. The semantic gap in multimedia IR remains a significant problem in the field, and solutions to them are many years off. However, new technological developments allow the use of knowledge organization and machine learning in multimedia search systems and services. Specifically, we argue that, the improvement of detection of some classes of low-level features in images music and video can be used in conjunction with knowledge organization to tag or label multimedia content for better retrieval performance. We provide an overview of the use of knowledge organization schemes in machine learning and make recommendations to information professionals on the use of this technology with knowledge organization techniques to solve multimedia IR problems. We introduce a five-step process model that extracts features from multimedia objects (Step 1) from both knowledge organization (Step 1a) and machine learning (Step 1b), merging them together (Step 2) to create an index of those multimedia objects (Step 3). We also overview further steps in creating an application to utilize the multimedia objects (Step 4) and maintaining and updating the database of features on those objects (Step 5).

Received: 27 August 2019; Revised: 14 October 2019, 8 November 2019; Accepted: 15 November 2019

Keywords: features, machine learning, knowledge organization, multimedia

† Many thanks to Sven Bale for his advice and clarification of features in music.


A. MacFarlane, S. Missaoui and S. Frankowska-Takhari. On Machine Learning and Knowledge Organization … 46

1.0 Introduction AI techniques, in particular machine learning have become a significant technology in information retrieval software and services. Machine learning is defined as a method that learns from data with minimal input from humans. A key example is search engines (Dai et al. 2011), which uses learning to rank algorithms to keep results presentation up to date given the inherent dynamism of the web. The web changes constantly both in terms of content and user re-quests, the data being documents, queries and click throughs, etc. For text retrieval, the machine learning in-frastructure is an essential part of the provision of a ser-vice that meets user needs, and there is a large body of research for this domain going back many years (Smiraglia and Cai 2017). The same however, could not be said of multimedia information retrieval where many challenges are still evident, although technological developments are beginning to change the situation. By multimedia retrieval we mean search for non-text objects such as images, pieces of music (Byrd and Crawford 2002) or videos/moving im-ages (Hu et al. 2011). Because of the semantic gap (Enser 2008), the features of these objects can be hard to identify and index, which leads to a separation of techniques in terms of concept-based retrieval and content-based re-trieval (with text we have terms that represent both). In MacFarlane (2016), it was argued that human involvement is necessary in many circumstances to identify concepts recognizable to humans—the example being a picture of a politician in an election. Whilst the politician can be easily identified (the “ofness” of the image), the election is a more nebulous concept that is difficult to extract from an image, without context (the “aboutness” of the image). Low-level features of objects are often difficult if not im-possible to match with concepts, and this problem is likely to be one that persists for a significant length of time. Knowledge organization methods are essential to ensure that these conceptual features are captured and recorded in multimedia software and services.

In this paper, we address the technological changes that have led to the potential for improvements in multimedia search and argue that knowledge organization can be used together with a supervised learning technique. We then re-view the landscape of multimedia search and show some possibilities for using knowledge organization and ma-chine learning to improve results for users in some types of information needs. Features in various types of multi-media objects are reviewed and we provide some advice on how to use these features and machine learning in con-junction with knowledge organization in multimedia IR systems and services. We provide some ideas for the way forward together with the practical implications for knowledge organization practitioners. The contribution of

the paper is a process model that uses knowledge organi-zation schemes and machine learning algorithms to create a database of objects for the purposes of multimedia in-formation retrieval. The proposed process model uses both high-level and low-level features identified for a mul-timedia object and the creation of an index within a data-base for the purpose of retrieval. 2.0 Machine learning and technological develop-

ments for machine learning What are the key developments that have led to improve-ments in technology, and which have significant implica-tions for the use of knowledge organization in multimedia search? In recent years, deep learning has become much more prominent in machine learning circles (Pouyanfar et al. 2018) for a wide range of different applications such as speech processing and machine vision (Deng and Yu. 2014). As you would expect there is a wide range of defi-nitions of deep learning, depending on the context, but the most appropriate in this context is a “class of machine learning techniques that exploit many layers of non-linear information processing for supervised or unsupervised feature extraction and transformation, and for pattern analysis and classification” (Deng and Yu 2014).

Whilst the underlying technology for deep learning (ar-tificial neural networks) has been around for many years (McCulloch and Pitts 1943), it is only recently that the use of the techniques has become widespread and available in open frameworks such as TensorFlow (Abadi et al. 2016). Over the years, the AI community has developed a strong body of knowledge in the use of the techniques, but a key turning point has been the availability of graphical pro-cessing units (GPUs), which are specialist chips that are able to significantly increase the processing of arithmetical operations (Singer 2019). They are particularly useful for image processing but have become very useful generally for other types of applications such as neural networks that require significant processing of numbers.

A benchmarking experiment conducted by Cullinan et al (2013), showed significant advantages for the GPU over CPU’s (central processing units) in terms of raw pro-cessing. The raw processing power from GPUs has proved to be the catalyst for a massive increase in the deployment of deep learning algorithms, in areas such as machine vi-sion to detect features in images. This includes features such as the detection of neuronal membranes (Ciresan et at. 2012), breast cancer (Ciresan et at. 2013) and handwrit-ten Chinese character recognition (Ciresan and Meier 2015).

Such advances in machine learning methods, including machine vision algorithms (Karpathy and Li 2015), have provided the functionality to identify specific objects in



images giving multimedia IR designers and implementers the ability to address the semantic gap to some extent. It is argued that in conjunction with knowledge organization, machine learning can be used to provide better and more relevant results to users for a given set of needs that re-quire the identification of specific objects to resolve or ad-dress that information need. This paper puts forward an argument for a supervised learning approach in multime-dia search, where a knowledge organization scheme is used as a rich source of information to augment the objects identified by any machine learning algorithm. This is to provide an enhanced index of objects, allowing more ef-fective search for those objects by the user. We review the overall approach we advocate when using machine learn-ing in conjunction with knowledge organization next. 3.0 Machine learning and knowledge organization As feature extraction from various media has improved in recent years through developments overviewed in Section 2, what are the implications for the use of knowledge or-ganization techniques? Knowledge organization in its many forms (thesauri, taxonomies, ontologies) are human generated schemes, which provide a rich source of evi-dence to describe features of objects that are of interest—in this case, multimedia objects such as images, music and video. The key to understanding the contribution knowledge organization can make in multimedia search is to consider the types of learning: unsupervised, semi-su-pervised and supervised (Russell and Norvig 2016). These are classed by their access to labelled or categorised data. Unsupervised learning (Russell and Norvig 2016, 694) is where algorithms work without any labelled data, for ex-ample, with clustering objects together based on the fea-tures extracted from them. This does not apply to our con-text, where we examine the use of knowledge organization techniques to the problem. Semi-supervised learning (Rus-sell and Norvig 2016, 695) does have some access to some labelled data, and it is possible to use this technique in some contexts where a limited number of multimedia ob-jects have been manually classified by a practitioner. Super-vised learning (Russell and Norvig 2016, 695) requires ac-cess to data that is completely labelled and is appropriate here—where we consider a large number of multimedia objects have been classified by a practitioner. We can either match features detected by both the machine learning al-gorithm and the practitioner (exact match case) or estimate the probability of a features matching from both sources using supervised learning techniques (best match case). We consider both examples later on the paper in Section 6. In this paper, we focus on the user of knowledge organiza-tion and supervised learning in multimedia search, in the context of large amounts of data that have been labelled

by practitioners. The scope of our work is in the use of non-symbolic AI methods (such as neural networks), ra-ther than symbolic methods deployed in prior work when knowledge organization has been used with machine learn-ing, e.g., in expert systems (Lopez-Suarez and Kamel 1994). 4.0 Machine learning and multimedia information

retrieval There are limits to the use of machine learning/AI tech-niques to the application of multimedia information re-trieval (MacFarlane 2016). However, new advances in tech-nology laid out in Section 2 above and the ability of ma-chine learning algorithms to detect objects in media, e.g. images (Karpathy and Li 2015), have provided scope to improve multimedia search results using knowledge organ-ization . In MacFarlane (2016), we argue that media of var-ious kinds (e.g., images, music) requires cultural knowledge that can often be only expressed tacitly and require human input. The advantage of knowledge organization schemes is that they provide this knowledge that is hard for machine learning algorithms to detect and can, therefore, be used with features extracted from multimedia objects to aug-ment the indexing of that object.

The key to understanding the application of knowledge organization and machine learning to multimedia infor-mation retrieval problems is to consider different types of information needs in particular domains. One particular domain that provides useful examples is the creative do-main, where various media is required on a daily basis, e.g. video, music (Inskip et al. 2012) and images (Konkova et al. 2016), for advertising campaigns, images for online news stories (Frankowska-Takhari et al. 2017). A specific example of information needs is the use of briefs in the advertising world, which provide an overview of the media required and some specification of the criteria for the ob-ject to be suitable for that particular campaign. Analysis of these briefs has demonstrated that there are some aspects that can be easily detected by machine learning algorithms, whilst others are too abstract for current techniques to work. For example, in music, Inskip et al. (2012) found that mood was a significant criterion for relevance in music briefs, which would be hard for an algorithm to detect. However, knowledge organization schemes with human input can help to resolve the need. Inskip et al. (2012) also found that music features such as structure are also im-portant, which machine learning algorithms can clearly be applied to. In terms of images, Konkova et al. (2016) found three categories of facets in image briefs including syntactic feature such as “colour” and “texture” as 'well as high-level general and conceptual image features such as “glamorous” and “natural.” These aesthetic features are an



open problem in the field (Datta et al. 2008). As with mu-sic, there is a clear distinction as to which image facets can be detected using machine learning algorithms.

Machine learning algorithms are very often used to de-tect features (Datta et al. 2008) in a variety of different ap-plications. The full range of algorithms can be found in Datta et al. (2008), Pouyanfar et al. (2018) and Murthy and Koolagudi (2018), but what problems are the algorithms applied to in the context of multimedia IR? Key problems that are addressed in many applications are classification, object detection and annotation. Examples include images where superhuman performance has been recorded in the 2015 large scale visual recognition challenge (ILSVRC’15) using deep learning methods (Poyyanfar et al. 2018), which has come about due to much improved object recognition (improving the ability to detect objects improves classifi-cation techniques). This has also led to techniques that can automatically annotate and tag images, including online services such as Imagga (https://imagga.com/). In music, techniques to apply classification and temporal annotation have been developed at low-level (e.g., timbre), mid-level (e.g., pitch and rhythm) and high-level (e.g., artist and genre) in many music applications (Murthy and Koolagudi 2018). In video (which is moving images together with sound), problems addressed include event detection by lo-cating scene changes and segmentation of the object into stories, e.g., scenes and threads in a TV programme or film (Lew et al, 2006). A quick review of the literature shows that machine learning has been applied to many problems in multi-media successfully, but there are many issues to which the technique cannot be addressed (see above). The key, therefore, to augmenting any application that uses knowledge organization as its core with machine learning, is to identify the features to which the technique can be used. It is these features that have been used successfully in the field that are known to bear fruit given the empirical evidence available. It is to these that we turn to next. 5.0 Features in multimedia information retrieval Features are aspects of an object that can be used for mul-timedia search purposes. The key to the application of search on multimedia objects is to identify these features and provide an index for them, allowing for applications such as direct search and classification or categorisation. In this section, we review the features for images, music and video and provide an overview of what machine learn-ing can identify and what is appropriate for knowledge or-ganization techniques and when both can be combined. Our emphasis is on combining the features from both sources to improve multimedia search applications and services.

5.1 Image features There is a wide variety of schemes that identify image at-tributes for modelling image retrieval. These include se-mantic (e.g., Panofsky/Shatford), syntactic and non-visual attributes (Westman 2009, 65-66). While non-visual attrib-utes (such as the meta-data, e.g. bibliographic data) can be useful (Konkova et al. 2016), this is not the concern here, as we focus on the semantic and syntactic features. One of the earliest frameworks is Panofsky’s theory (Panofsky, 1962) that describes three levels of meaning in a work of art: pre-iconographical, iconographical and iconological. Shatford (1986) extended this model and proposed that se-mantic information in an image may be analysed on the level of generic and specific elements present in the image (the “ofness” of the image), and on the level of the ab-stract themes present in the image (the “aboutness” of the image). While describing the “ofness” involves decoding and naming of the objects in the image, interpreting the “aboutness” from the image, especially, an image rich in symbolic meaning (e.g., a work of art), requires previous personal, cultural knowledge and experience from the viewers. Therefore, semantic information for an image will require human input to establish the “aboutness” of a given object. Currently, this can be done through generic schemes such as the Thesauri for Graphic Materials (Li-brary of Congress N.D.b), and specific schemes such as Iconclass (http://www.iconclass.nl/) that is focused on art images. While most existing frameworks stem from the Panofsky/Shatford matrix (Shatford 1986), the more re-cent models (e.g., Eakins et al. 2004; Hollink et al. 2004; Jaimes and Chang 2000) allow the distinction between the semantics and syntax of images. Syntactic attributes can either be primitive visual elements such as colour, texture, hue and shape, or compositional, e.g., relationship between shapes, motion, orientation, perspective, focal point (West-man 2009, 65).

It is these syntactic attributes to which machine learning can be applied. Specific application areas have particular needs. For example, the concept of “copyspace” is im-portant in advertising, which is a clear space to insert text (Konkova et al. 2016). Further, studies from the user-cen-tred tradition advocate that human image users in specific domains have specific image needs. Such studies aim to uncover the needs of users and identify which aspects of user needs can be used to facilitate automation of image-based tasks. For example, Frankowska-Takhari et al. (2017) investigated the needs of image users in online journalism. Initially, their findings were similar to those from earlier studies, e.g., Markkula and Sormunen (2000), Westman and Oittinen (2006), and showed that users’ descriptions of their image needs were often limited to their conceptual needs, and search queries tend to relate to concepts, while



information about users’ needs on the perceptual level was limited to descriptions of visual effects required in images.

As suggested in Machin and Polzer (2015), it was neces-sary to reach beyond these descriptions, to identify the con-crete visual features that engendered the required effects. Frankowska-Takhari et al. (2017) applied the visual social se-miotics framework (Kress and van Leeuwen, 2006) to ana-lyse images used in online journalism. They identified a set of eleven recurring visual features that engender the visual effect required in images used for illustrating news headline content (see Table 1). These included: a strong single focal point to draw readers’ attention, the use of specific palette of colours depending on the tone of the news story, a pho-tographic shot from waist-up including head and shoulders and close-up on the face, and a preference for a large ob-ject/person in the frame. Most of the identified features are detectable to currently available systems that make use of advanced computer vision. They could be implemented, for example, as multi-feature filters for image retrieval. Such a system firmly rooted in the image users’ needs, could be a step towards automating image retrieval with a purpose to support a specific group of image users carrying out specific illustration tasks. 5.2 Music features Downie (2002) identifies seven facets of music information that can be considered as features to learn for a retrieval system, which can be further classified into low-level, mid-level and high-level features (Murthy and Koolagudi 2018). We merge these two schemes together as they provide a useful overall classification of features in which machine learning can be applied and where knowledge organization schemes are appropriate, as well as identifying the key fea-

Feature Visual image features

1 The specific (identifiable) person/people related to the topic depicted in the image

2 The person/people depicted in the foreground

3 Shot from waist up

4 Face visible: frontal or profile shot

5 Gaze: direct or side gaze

6 The depicted person is “large” in the frame

7 Positioned centrally or to the right within the frame

8 Colour image

9 Colour intensity: saturated or soft colours used

10 Blurry or monotone background

11 The person’s face in focus (sharp)

Table 1. Image features recurring in news headline images. Source: Frankowska-Takhari et al. (2017).

tures. The features are not mutually exclusive (Downie 2002), and low-level features are used to build mid-level fea-tures, which in turn can be used to extract high-level fea-tures (Murthy and Koolagudi 2018). Low-level features are defined as the fundamental property of sound, mid-level features the fundamental properties of music and high-level features the human perceptual interpretation of the mid-level features.

The low-level features are timbre and tempo. Timbe is defined as an attribute related to the tone, that differs in the instrument being played (e.g., trumpet vs piano). It is the sound, tone quality and colour that make up the voice qual-ity of a musical note (Murthy and Koolagudi 2018, 7). Tempo is defined as the duration between two musical events (e.g., two notes). Timbre and tempo are strongly con-nected through frames, a short time segment of 10-100ms. These low-level features can fail to capture much infor-mation from a given song in their own right (Murthy and Koolagudi 2018) and mid-level features are required to build up a picture of music that can be used for an application. These mid-level musical features are pitch, rhythm, har-mony and melody—note that in our scheme these features are still low-level. Pitch is frequency of sound, the oscilla-tions per second. Differences between two pitches are de-fined as being the interval between them. Harmony is de-tected when two or more pitches sound at the same time to create polyphonic sound, which is determined by the inter-val. Rhythm is defined by an occurring or recurring pattern in the music, e.g., the beat. Rhythm and pitch determine a further important feature of music namely melody, which is a succession of musical notes. Murthy and Koolagudi (2018) do not classify this feature, but it is clearly a mid-level feature as it strongly related to other mid-level features but cannot be regarded as a high-level feature. It is these mid-level fea-tures to which machine learning can be applied.

There is more ambiguity in terms of high-level features and some can be detected through learning mid-level fea-tures, but others require human input. In some, both ma-chine learning and knowledge organization can be used. High-level features include editing, text, bibliography (Downie 2002) and artist, genre, instrument and emotion (Murthy and Koolagudi 2018). Editing is defined as perfor-mance instructions of a piece of music such as fingering, articulation, etc. Knowledge organization schemes such as the Library of Congress performance terms for music (Li-brary of Congress 2013c; 2013d) focused largely on western classical music, are appropriate. Text relates to any lyrics as-sociated with a musical piece and can be handled via normal text retrieval techniques. It may be appropriate to use this feature to augment machine learning algorithms (in con-junction with natural language processing techniques). Bib-liography refers to the meta-data of the piece, which is de-



termined by human entry of aspects such as composer, per-former, etc. Appropriate meta-data standards in the field are applied here, and as with text can be used to augment ma-chine learning algorithms. Bibliography can determine the artist, genre, emotion and instrument features (depending on the meta-data scheme used), but machine learning has been used to identify those high-level features from mid-level features extracted from a musical piece, e.g., to classify it by the given feature (Murthy and Koolagudi 2018). The genre feature can also be augmented with knowledge organ-ization schemes such as the Library of Congress mu-sic/genre headings (2013a; 2013b). 5.3 Video features Video is multimedia in the complete sense as it consists of moving images in sequence with audio. Image features identified in 5.1 above can be used here, and as we have extra evidence (e.g., a series of images) we have more evi-dence to improve the detection of objects in the media being indexed. A practical example of the features that can be identified are outdoor and indoor shots, people and landscapes/cityscapes (Smeaton and Over 2002). There are many features from audio that can be extracted via ma-chine learning including speech to text (where text retrieval techniques can be used) and music (see 5.2 above). Whilst we can build on these features, there are unique features of video that can be used to classify or segment video ob-jects. Video can be split up into scenes and threads (Lew et al. 2006), for example in a news programme where dif-ferent news stories are presented to the viewer. The TRECVID track at the TREC (Text retrieval Conference) investigated this in the shot boundary detection task (Smeaton and Over 2002) by detecting different categories, e.g. cut (sort finishes, one starts right after), dissolve (one shot fades out while new one fades in), fadeout/in (one shot fades out then new one fades in) plus other categories which don’t fit into these precise boundaries. Detecting shot boundary allows the detection of higher-level features such as events, embodied in LSCOM (http://www.ee.co-lumbia.edu/ln/dvmm/lscom/), the large-scale concept ontology for multimedia (Naphade et al. 2006). This is a knowledge organization scheme built via the empirical work carried out by the multimedia community, with TRECVID being particularly notable. Examples include people crying (007), maps (204) and people associated with commercial activities (711). These features can be aug-mented with other knowledge organization schemes such as the Library of Congress (N.D.a) scheme for assigning genre/form terms to films and video.

5.4 Summary of features In this section, we have identified two classes of features, one to which machine learning can be applied and one which cannot. The low-level features such as colour and hue in images, pitch and tempo in music and shot bound-aries in video are ones that can be extracted using machine learning techniques, whilst high-level features such as “aboutness” require the use of human intervention via the application of knowledge organization schemes. Next, we consider the use of these different classes of features in conjunction with each other to improve multimedia infor-mation retrieval services. 6.0 Using machine learning and knowledge organi-

zation to enhance multimedia information retrieval

We propose a process model by which the features for a multimedia object are identified (both high-level and low-level) to create a database of objects for the purposes of retrieval. We assume access to digital objects (analogue ob-jects are not considered here). We identify five steps in this process model (see Figure 1). In Step 1, we identify the corpus and knowledge organization scheme for the given corpus, which is split into two separate sub-steps: applying the knowledge organization scheme to the high-level cor-pus objects (1a) and using machine learning to identify the low-level object features (1b). In Step 2, we combine both high and low-level object features to provide a comprehen-sive set of features for multimedia, which is richer for re-trieval purposes (Step 3). From Step 3 we have the infor-mation to create the application of our choice, either a classification or categorization system, or to support mul-timedia search functionality (Step 4). A further Step is con-sidered (Step 5), given two scenarios—either a new set of features is identified (by a change in the knowledge organ-ization scheme or improved feature detection using ma-chine learning) or a new set of objects is received and needs to be indexed. We discuss each of these Steps below, highlighting the input and output data for each Step. 6.1 Step 1a: apply knowledge organization scheme

to corpus

Input Data Output Data

1. Corpus 2. Knowledge Organization

Scheme

Object features (high-level)

Table 2. Data required for Step 1a.

The information professional needs to choose a relevant knowledge organization scheme for the corpus they are



managing. This will either be a standard scheme (examples are cited in Section 5 above), or a specialist in house scheme derived by the organization that requires access to the multimedia. Collection size is a concern here—unless there are significant human resources, manually catalogu-ing multimedia objects using the knowledge organization scheme might not be practical. In this case any meta-data associated with the object can be used, with knowledge or ganization applied to the meta-data to identify relevant fea-tures for the database. In other cases, the corpus will al-ready have been indexed (perhaps over many years) and high-level features for each object will be readily available. If the media contains speech (if the corpus is either audio or video that contains audio), machine learning can be used to detect text, on which the knowledge organization can be applied. Whilst the word error rates might be high, the main bulk of concepts for the objects will be detected. This text might itself be indexed as part of the multimedia search service.

An example to illustrate this is from the advertising do-main. Konkova et al. (2016) provide a list of facets for im-ages in which knowledge organization elements can be placed. Examples of this are image style and conceptual features, which are very subjective and require human in-put. Image style could include glamour, whether it is natu-ral or manipulated (using photoshop), amateur or profes-sionally taken photo. Conceptual features could include positive busy images of bustling street life, inno-cence/guilt, freedom/slavery, beauty/ugliness, etc (the “aboutness”). General semantics of what is in the image could also be detected, e.g., beautiful images of clouds on the planet Jupiter, family walking together on a beach, etc.

6.2 Step 1b: apply machine learning technique to corpus


Corpus Object features (low-level)

Table 3. Data required for Step 1b.

The next step for any information professional is to iden-tify the low-level features using machine learning. This may require the assistance of technical staff with AI expertise, but the information professional should be aware of the process used to generate these features. A key decision is to identify training and test objects from the corpus or a subset of the corpus. The training set is used to detect the features from the corpus, whilst the test set is used to val-idate the features detected. Getting this right is key, as poor decisions can lead to over fitting of features, reducing their utility for retrieval purposes. In general, the standard way to split the corpus into training and test collections is two thirds for training and one third for testing at least. The training set should always be much larger than the test set. A further step is to split a corpus into a number of seg-ments (say k) and spilt each of these k segments applying the machine learning algorithms to each of these seg-ments, by treating each k segment as a test set with other segments as the training set. This can be repeated with all of the segments and the results merged to create a set of features that is more robust. This is known as cross-vali-dation.

The type and size of corpus is a consideration. The pro-fessional should consider appropriate features identified in

Figure 1. Process using knowledge organization and machine learning to index multimedia.



Section 5 for their corpus, and the training and test sets should not be too large (in some cases corpuses with many millions of objects and large features sets may be difficult to manage as machine learning is computationally inten-sive). It should be noted that in order to get an unbiased estimate of how good your algorithm is doing, it was com-mon practice to take all your data and split it according to a 70/30% ratio (i.e., 70/30 train test splits explained above). These ratios were perfectly applied when dealing with small datasets. However, in the big data and deep learning era, where, the data could exceed millions of in-stances, the test sets have been becoming a much smaller percentage of the total. For example, if you have a million examples in the dataset, a ratio of 1% of one million or so (99% train, 1% test) will be enough in order to evaluate your machine learning algorithm and give you a good esti-mate of how well it’s doing. This scheme is manageable for large datasets. However, any sample chosen must also be representative, otherwise the features will not be valid. At the end of this step, the low-level object features will be identified.

An example to illustrate this is from the advertising do-main. Konkova et al. (2016) identifies a list of facets ripe for the application of machine learning. Composition of the image can be detected such as shooting distance (close up, panoramic view of a landscape), angle (shot taken from the left of a subject), object location (lamp on a desk) or focus (sharp, blurred). Light is a related facet where the time of day can be detected (shadows), type of light (nat-ural, artificial) and by location (outside or inside shot). Spe-cific semantics including particular entities/places/people can be detected, e.g., a human hand holding an archaeo-logical artifact, a shot of St Peters Basilica in Rome, etc. 6.3 Step 2: merge features for multimedia objects


1. Object features (high-level) 2. Object features (low-level)

Object features (combined)

Table 4. Data required for Step 2.

The data produced in Step 1 from both sub-steps needs to be merged together to create a comprehensive set of fea-tures for each object in the multimedia corpus. It is this comprehensive set of features that provides the enhance-ment required for better multimedia retrieval. Getting the merge process correctly configured is, therefore, critical, and there are two cases to consider: one straightforward and one that requires a little more thought. The simpler case is the exact match case, split into conjoint and disjoint sub-cases. In conjoint sub-case, we have the same feature identified in both inputs (e.g., text extracted from images

may match a term in the knowledge organization scheme) and record that feature in the index. In most cases, the fea-tures will be distinct (the disjoint sub-case—a feature is identified either by the knowledge organization scheme OR by the machine learning algorithm) and the infor-mation professional will need to think about which fea-tures to record. They may think it appropriate to record all features, but this may have drawbacks (features may not be useful for search). One way to get around this is to use machine learning to see which of the low and high-level features correlate with each other in the input dataset and choose the best set of features—this is the best match ap-proach. In this, either all inputs from both sources or from the disjoint sub-case could be used. This would work by applying a further step of machine learning (as outlined in Step 1b above), in which an appropriate sample would be used to generate a set of features for indexing. The advice given in Section 6.2 would apply in the best match case. At the end of this, a full set of features appropriate for search will be identified. There are many different contexts to consider, and the information professional will need to be clear about the particular implications for their given situ-ation.

Taking the example given from the advertising domain above (Konkova et al. 2016), this would appear to be dis-joint and the features about any given image object can be merged together quickly and easily. The facets and their qualities are really quite different and distinct, and it is clear which process will create the appropriate image descrip-tion for that facet. It should be noted that improvements in machine learning may address the general semantics facet, which may need reviewing by the image indexer. 6.4 Step 3: create index of features

(database of objects)


Object features (combined) Database of Objects (Index)


Once a full set of features has been identified, an index of objects using those features can be generated. This can be either an inverted list or a relational or object relational da-tabase, depending on the context. The information profes-sional could consult a technical person to assist with this. Examples of software available include Elasticsearch (https://www.elastic.co/), MongoDb (https://www.mon-godb.com/), Neo4j (https://neo4j.com/), MySQL (https:// www.mysql.com/) and PostgresQL (https://www.post gresql.org/).



6.5 Step 4: create application or service with combined features


Link to Database of Ob-jects (Index)

Object classification or cate-gorisation


Once the database has been created, the application or ser-vice to meet user needs can be produced. For retrieval pur-poses, this may just mean writing an appropriate front end given users’ needs, together with a back end that matches user defined features identified at the front end. However, if categorisation or classification were required, a further round of machine learning would be appropriate. This would be taking the machine learning process overviewed in Step 1b above but applying the algorithm to the com-bined feature set. An example can be found in Fan et al (2007), who combined wordNet and ontology data to sup-port a surgery education application. 6.6 Step 5: Update database of objects with new in-

formation


1. New Objects 2. New Features

1. Updated Database 2. Updated Features and Database

Table 7. Data required for Step 5. New information is generated all the time, and an infor-mation professional cannot assume that the corpus they manage will remain static. There are two scenarios to con-sider—one where new multimedia objects are received and need to be considered and one were new features are avail-able. The first of these is easy to deal with as features can be assigned (high-level features in the knowledge organi-zation scheme, low-level features extracted by an algo-rithm) and the object recorded in the database. The second is not so straight forward and it requires a restart of the process—either because new elements have been added to the knowledge organization scheme or because machine learning algorithms have been improved to provide a clearer picture of a feature already identified or to identify new features. This will be an expensive and time-consum-ing process, so the information professional may wish to test the ideas on a sub-set of the corpus before restarting the whole process again. 7.0 Conclusion In this paper, we put forward some practical advice for in-formation professionals who curate multimedia digital col-

lections and who are charged with supporting search ser-vices to those collections. We believe that information pro-fessionals should treat machine learning and/or AI tech-niques an opportunity rather than a threat and should se-riously think about using technology to improve the mul-timedia services they manage. Information professionals should be wary of the hype that surrounds machine learn-ing/AI that has all too often been overhyped in terms of impact, leading to AI winters. However, the process model we describe in Section 6 we believe gives the information professional an opportunity to seize the initiative and build on their domain knowledge gained in working on images, music and video. We urge the community to consider this when considering access to multimedia digital objects for their users. References Abadi, Martin. Paul Barham, Jianmin Chen, Zhifeng Chen

andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, Manjunath Kudlur, Josh Levenberg, Rajat Monga, Sherry Moore, Derek G. Murray, Benoit Steiner, Paul Tucker, Vigay Vasudevan, Pete Warden, Martin Wicke, Yuan Yu and Xiaoqiang Zheng. 2016. “Tensorflow: A System for Large-Scale Machine Learning.” In Proceedings of OSDI ’16: 12th USENIX Symposium on Operating Systems Design and Implementation November 2–4, 2016 Savannah, GA, USA, ed. Andrea Arpaci-Dusseau and Geoff Voelker. 16. Carlsbad, CA: Usenix, 265-83.

Byrd, Donald and Tim Crawford. 2002. “Problems of Mu-sic Information Retrieval in the Real World.” Information Processing & Management 38: 249-72.

Cireşan, Dan C., Alessandro Giusti, Luca M. Gambardella and Jurgen Schmidhuber. 2012. “Deep Neural Net-works Segment Neuronal Membranes in Electron Mi-croscopy Images.” In Advances in Neural Information Pro-cessing Systems 25: 25th Annual Conference on Neural Infor-mation Processing Systems 2011, December 12-15, 2011, Gra-nada, Spain, ed. Fernando Pereira, Chris Burges, Leon Bottou and Kilian Weinberger. La Jolla, CA: Neural In-formation Processing Systems; Red Hook, NY: Printed from e-media with permission by Curran Associates, 2843-51.

Cireşan, Dan C. Alessandro Giusti, Luca M. Gambardella and Jurgen Schmidhuber. 2013. “Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks.” In International Conference on Medical Image Computing and Computer-Assisted Intervention, ed. Kensaku Mori, Ichiro Sakuma, Yoshinobu Sato, Christian Baril-lot and Nassir Navab. 16. Lecture Notes in Computer Science 8150. Berlin: Springer-Verlag, 411-8.



Cireşan, Dan C. and Ueli Meier. 2015.“Multi-Column Deep Neural Networks for Offline Handwritten Chi-nese Character Classification.” In International Joint Con-ference on Neural Networks, ed. Yoonsuck Choe. New York: IEEE. doi:10.1109/IJCNN.2015.7280516

Cullinan, Christopher, Christopher Wyant and Timothy Frattesi. 2019. “Computing Performance Benchmarks among CPU, GPU and FPGA.” Accessed November 8. http://www.wpi.edu/Pubs/E-project/Available/E-pro

ject-030212-123508/unrestricted/Benchmarking_Final. pdf Dai, Na., Milda Shokouhi and Brian D. Davison. 2011.

“Learning to Rank for Freshness and Relevance.” In Pro-ceedings of the 34th International ACM SIGIR Conference on Research and Development in Information Retrieval, ed. Richardo Baeza-Yates, Tat-Seng Chua and W. Bruce Croft. 34. New York: ACM, 95-104. doi:10.1145/200991 6.2009933

Datta, Ritendra, Dhiraj Joshi, Jia Li and James Z. Wang. 2008. “Image Retrieval: Ideas, Influences and Trends of the New Age.” ACM Computing Surveys 40, no. 2: article no. 5. doi:10.1145/1348246.1348248

Deng, Li and Dong Yu. 2014. Deep Learning: Methods and Applications. Foundations and Trends in Signal Processing. Delft: Now.

Downie, J. Stephen. 2003. “Music Information Retrieval.” Annual Review of Information Science and Technology 37: 295-340.

Eakins, John P., Pam Briggs and Bryan Burford B. 2004. “Image Retrieval Interfaces: A User Perspective.” In: Proceedings of the 3rd International Conference on Image and Video Retrieval. ed. Peter Enser, Yiannis Kompatsiaris, Noel E. O’Connor, Alan F. Smeaton and Arnold W. M. Smeulders. Lecture Notes in Computer Science 3115. Berlin: Springer, 628-37.

Enser, Peter G.B. 2008. “The Evolution of Visual Infor-mation Retrieval.” Journal of Information Science, 34: 531-46.

Fan, Jianping, Hangzai Luo, Yuli Gao and Ramesh Jain. 2007. “Incorporating Concept Ontology for Hierar-chical Video Classification, Annotation and Visualiza-tion.” IEEE Transactions on Multimedia 9: 939-57.

Frankowska-Takhari, Sylwia. Andrew MacFarlane, Ayse Göker and Simone Stumpf. 2017. “Selecting and Tailor-ing of Images for Visual Impact in Online Journalism.” Information Research 22, no. 1. http://informationr.net/ ir/22-1/colis/colis1619.html

Hollink, Laura, Guss Schreiber, Bob J. Wielinga and Mar-cel Worring. 2004. “Classification of User Image De-scriptions.” International Journal of Human Computer Stud-ies 61: 601-26.

Hu, Weiming, Nianhua Xie, Li Li, Xianglin Zeng and Ste-phen Maybank. 2011. “A Survey on Visual Content-Based Video Indexing and Retrieval.” IEEE Transactions

on Systems, Man and Cybernetics, Part C (Applications and Reviews) 41: 797-819.

Inskip, Charlie. Andrew Macfarlane and Pauline Rafferty. 2012. “Towards the Disintermediation of Creative Music Search: Analysing Queries to Determine Important Fac-ets.” International Journal on Digital Libraries 12: 137-47.

Karpathy Andrej and Li Fei-Fei. 2015. “Deep Visual-Se-mantic Alignments for Generating Image Descriptions.” In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2015, ed. Kristen Grauman, Eric Learned-Miller, Antonio Torralba and Andrew Zisser-man. Boston: IEEE, 3128-37. https://www.cv-founda

tion.org/openaccess/content_cvpr_2015/papers/Kar pathy_Deep_Visual-Semantic_Alignments_2015_CV PR_paper.pdf Konkova, Elena. Andrew MacFarlane and Ayse Göker.

2016. “Analysing Creative Image Search Information Needs.” Knowledge Organization 43: 14-21.

Jaimes, Alejandro and Shih-Fu Chang. 2000. “A Conceptual Framework for Indexing Visual Information at Multiple Levels.” In Proceedings Volume 3964 Electronic Imaging | 22-28 January 2000 Internet Imaging, ed. Giordano B. Beretta and Raimondo Schettini. 396. doi:10.1117/12.373443

Kress, Gunter and T. van Leeuwen. 2006. Reading Images: The Grammar of Visual Design. London: Routledge.

Lew, Michael S., Nicu Sebe. Chabane Djeraba and Ramesh Jain. 2006. “Content-Based Multimedia Information Retrieval: State of the Art and Challenges.” ACM Trans-actions on Multimedia Computing, Communications and Appli-cations (TOMM) 2: 1-19. doi:10.1145/1126004.1126005

Library of Congress. 2013a. “Genre/Form Terms for Mu-sical Works and Medium of Performance Thesaurus.” https://www.loc.gov/catdir/cpso/genremusic.html

Library of Congress. 2013b. “Genre/Form Terms Agreed on by the Library of Congress and the Music Library Association as in Scope for Library of Congress Genre/ Form Terms for Library and Archival Materials (LCGFT).” http://www.loc.gov/catdir/cpso/lcmlalist.

pdf Library of Congress. 2013c. “Introduction to Library of

Congress Medium of Performance Thesaurus for Mu-sic.” http://www.loc.gov/aba/publications/FreeLCSH/ mptintro.pdf

Library of Congress. 2013d. “Performance Terms: Me-dium.” http://www.loc.gov/aba/publications/FreeLCS H/MEDIUM.pdf

Library of Congress. 2019a “Library of Congress Genre/ Forms for Films Video.” Accessed November 8. http:// www.loc.gov/aba/publications/FreeLCSH/GENRE.

pdf Library of Congress. 2019b “Thesaurus for Graphical Ma-

terials (TGM).” Accessed November 8. http://www.loc. gov/pictures/collection/tgm/



Lopez-Suarez, Alex. and Mohammed S. Kamel. 1994. “Dykor: a Method for Generating the Content of Ex-planations in Knowledge Systems.” Knowledge-Based Sys-tems 7: 177-88.

MacFarlane, Andrew. 2016. “Knowledge Organization and its Role in Multimedia Information Retrieval.” Knowledge Organization 43: 180-3.

Machin, David and Lydia Polzer. 2015. Visual Journalism. Journalism: Reflections and Practice. London: Palgrave.

Markkula, Marjo and Eero Sormunen. 2000. “End-User Searching Challenges Indexing Practices in the Digital Newspaper Photo Archive.” Information Retrieval 1: 259-85.

McCulloch, Warren S. and Walter Pitts. 1943. “A Logical Calculus of the Ideas Immanent in Nervous Activity.” Bulletin of Mathematical Biophysics 5: 115-33.

Murthy, Y.V. Srinivasa and Shashidhar G. Koolagudi, 2018. “Content-Based Music Information Retrieval (CB-MIR) and its Applications toward the Music Industry: A Review.” ACM Computing Surveys 51, no. 3: article no. 45. doi:10.1145/3177849

Naphade, Milind. John R. Smith, Jelena Tesic, Shih-Fu Chang, Winston Hsu. Lyndon Kennedy, Alexander Hauptmann and Jon Curtis. 2006. “Large-Scale Concept Ontology for Multimedia.” IEEE multimedia 13: 86-91.

Russell, Stuart J. and Peter Norvig. 2016. Artificial Intelli-gence: A Modern Approach. 3rd ed. [UK]: Pearson.

Panofsky, Erwin. 1962. Studies in Iconology: Humanistic Themes in the Art of the Renaissance. New York: Harper & Row.

Pouyanfar, Samira, Saad Sadiq, Yilin Yan, Haiman Tian, Yudong Tao, Maria Presa Reyes, Mei-Ling Shyu, Shu-

Ching Chen and S.S. Iyengar,. 2018. “A Survey on Deep Learning: Algorithms, Techniques and Applications.” ACM Computing Surveys 51, no. 5. doi: 10.1145/3234150

Shatford, Sara. 1986. “Analyzing the Subject of a Picture: A Theoretical Approach.” Cataloging and Classification Quarterly 6: 39-62.

Smeaton, Alan F. and Paul Over. 2003. The TREC-2002 Video Track Report. The Eleventh Text Retrieval Conference (TREC 2002), ed. Ellen Voorhees. NIST Special Publica-tion SP 500-251. Gaithersburg, MD: NIST. https:// trec.nist.gov/pubs/trec11/t11_proceedings.html

Smiraglia, Richard P. and Xin Cai. 2017. “Tracking the Evolution of Clustering, Machine Learning, Automatic Indexing and Automatic Classification in Knowledge Organization.” Knowledge Organization 44: 215-33.

Singer, Graham. 2019. “The History of the Modern Graphics Processor.” TechSpot (blog), November 21. https://www.techspot.com/article/650-history-of-

the-gpu/ Westman, Stina. 2009. “Image Users’ Needs and Searching

Behaviour.” In Information Retrieval in the 21st Century, ed. Ayse Goker and John Davies. Chichester: John Wiley & Sons, 63-83.

Westman, Stina and Pirkko Oittinen. 2006. “Image Re-trieval by End-Users and Intermediaries in a Journalistic Work Context.” Proceedings of the 1st IIiX Symposium on Information Interaction in Context, ed. Ian Ruthven. 1, New York: ACM, 102-10.


B. Włodarczyk. KABA Subject Headings and the National Library of Poland Descriptors in Light of Wojciech Wrzosek’s Theory …

56

KABA Subject Headings and the National Library of Poland Descriptors in Light of Wojciech Wrzosek’s

Theory of Historiographical Metaphors and Different Historiographical Traditions

Bartłomiej Włodarczyk

University of Warsaw, Faculty of Journalism, Information and Book Studies, 00-310 Warsaw, Poland, Bednarska 2/4, <[email protected]>

Bartłomiej Włodarczyk is an assistant professor in the Faculty of Journalism, Information, and Book Studies at the University of Warsaw. He received an MA in history (specialization in social history) and a PhD in library and information science from the University of Warsaw. His main research area is knowledge organization.

Włodarczyk, Bartłomiej. 2020. “KABA Subject Headings and the National Library of Poland Descriptors in Light of Wojciech Wrzosek’s Theory of Historiographical Metaphors and Different Historiographical Tradi-tions.” Knowledge Organization 47(1): 56-71. 60 references. DOI:10.5771/0943-7444-2020-1-56.

Abstract: The aims of this article are, first, to provide a necessary background to investigate the discipline of history from the knowledge organization (KO) perspective, and secondly, to present, on selected examples, a way of analyzing knowledge organization systems (KOSs) from the point of view of the theory of history. The study includes a literature review and epistemological analysis. It provides a preliminary analysis of history in two

selected universal Polish KOSs: KABA subject headings and the National Library of Poland Descriptors. The research is restricted to the high-level concept of historiographical metaphors coined by Wojciech Wrzosek and how they can be utilized in analyzing KOSs. The analysis of the structure of the KOSs and indexing practices of selected history books is performed. A particular emphasis is placed upon the requirements of classical and non-classical historiography in the context of KO. Although the knowledge about historiographical met-aphors given by Wrzosek can be helpful for the analysis and improvement of KOSs, it seems that their broad character can provide the creators only with some general guidelines. Historical research is multidimensional, which is why the general remarks presented in this article need to be supplemented with in-depth theoretical and empirical analyses of historiography.

Received: 19 August 2019; Revised: 2 December 2019; Accepted: 16 January 2020

Keywords: history domain, historiographical metaphors, research, domain analysis, historiography, knowledge organization

1.0 Introduction One of the basic goals of knowledge organization (KO) is to form the theoretical background to create new knowledge organization systems (KOSs) and improve exist-ing ones. These tools have different structures and contexts of use, but their primary purpose is to organize documents to provide users with a convenient way to access and man-age them as knowledge artifacts (Mazzocchi 2018). This means that KO is mainly based on the practical needs of researchers and others interested in gaining knowledge about a specific aspect of reality; but at the same time, it has to be deeply grounded in domain knowledge. It seems to be especially true in the context of domain-specific KOSs, where only detailed knowledge of a domain can form a sound foundation for all KO activities. The need for analysis of various domains is a fundamental task in contemporary

KO research (Smiraglia 2012), which emphasizes the differ-ences between underlying assumptions, paradigms, objects of inquiry, and methods used in different disciplines, sub-disciplines, and fields.

The object of analysis in this paper is the domain of history. The point of departure is the definition taken from the Oxford English Dictionary (https://www.oed.com/), which defines it as, “The branch of knowledge that deals with past events; the formal record or study of past events, esp. human affairs. Also: this as a subject of study.” His-tory is a separate discipline with a tradition dating back to antiquity. The works from this period such as the Chinese Spring and Autumn Annals with commentaries (Zuo Tradition 2016), Records of the Grand Historian by Sima Qian (1993), the Greek The Histories by Herodotus (1996), and History of the Peloponnesian War by Thucydides (2017) are the classic works of Eastern and Western civilizations. Since its in-


B. Włodarczyk. KABA Subject Headings and the National Library of Poland Descriptors in Light of Wojciech Wrzosek’s Theory … 57

ception in antiquity, the bulk of research has focused on political and diplomatic history. The state was stabilized in the nineteenth century with the professionalization of his-torical research (Finney 2005; Iggers 2005). Since the be-ginning of the twentieth century, and especially after the end of the Second World War, new research fields such as social history and historical anthropology have emerged.1 This turn was connected with new research methods. A work of a Polish historian Maria Nietyksza (1971) entitled Ludność Warszawy na przełomie XIX i XX wieku is a repre-sentative example of the social history approach. The au-thor analyzed the populace of the capital of Poland (then one of the cities of the Russian Empire), utilizing quanti-tative methods and a broad range of statistical sources. A different approach was taken by Anna Żarnowska (1985), who explored the life of workers in Warsaw at the turn of the nineteenth and twentieth centuries. The author used different primary sources than Nietyksza, biographical ma-terials, among others, and asked different research ques-tions. The subject of the work, to a large extent, was the workers’ culture and their daily life.

In order to provide users with useful KOSs in the disci-pline of history, there needs to be an understanding of this mix of ideas, methods, and approaches. The aims of this article are, first, to provide a necessary background to inves-tigate the discipline of history from a KO perspective, and secondly, to present, on selected examples, a way of analyz-ing KOSs from the point of view of a theory of history. The study makes use of a literature review and epistemolog-ical analysis. A starting point for the review was the Interna-tional Society for Knowledge Organization (ISKO) KO lit-erature database (https://www.isko.org/lit.html). Other sources, like Google Scholar (https://scholar.google.com/), were also used to find relevant literature. The main part of the paper is devoted to the use of epistemological theory developed by Polish historian Wojciech Wrzosek, who spe-cializes in the history and theory of modern historiography, to analyze KOSs. In this respect, the paper refers to Birger Hjørland’s epistemological approach (see 2002, 438-440). It also seems to be in line with the suggestion formulated by Richard P. Smiraglia (2015, 99), who pointed to specific work on impressionist artists, which, according to him, “could easily be viewed as a prelude to domain analysis of the French painting world.” Therefore, it appears that KO researchers should try to exploit domain knowledge, includ-ing different theories, to advance KO.

The objects of analysis are selected Polish thesauri and indexing practices followed by indexers from institutions that have developed these tools. The paper refers here to two approaches presented by Hjørland (2002), that is, con-structing special classifications and thesauri (see 2002, 425-428) and indexing and retrieving specialties (see 2002, 429-430). It should be stressed that the paper provides only a

preliminary analysis of history in two selected Polish KOSs: KABA subject headings (KABA) and the National Library of Poland Descriptors (NLPD). Furthermore, the analysis is mainly restricted to the high-level concept of historiographical metaphors and how they can be utilized in analyzing KOSs. The paper makes use of a specific his-toriographic framework as an example, but there are also other views that can be applied to the analysis of KOSs. Some of other authors working in the field of the theory and methodology of history are mentioned in the litera-ture review.

The remainder of the article is organized as follows: Section 2.0 presents a literature review with special consid-eration of domain-analytical works in history and related disciplines. Section 3.0 presents Wojciech Wrzosek’s con-cept of a historiographical metaphor as the theoretical background of research. The example of an epistemolog-ical analysis of two Polish KOSs from the perspective of two historiographical traditions and their historiographical metaphors is presented in Section 4.0. An analysis is per-formed on the structure of the KOSs and indexing prac-tices of selected history books. A special emphasis is placed upon the requirements of classical and non-classi-cal historiography in the context of KO. Lastly, a summary is presented. 2.0 Literature review A sizeable body of research exists in KO on different do-mains, but only a small number of studies have focused on history. This section provides a review of the literature about domain analysis in general, along with selected pa-pers and books related to the use of different approaches in KO and information science to the domain of history and related disciplines. Additionally, some chosen works about history, interesting from the perspective of the goal of this paper, are also presented. This part can be deemed an introduction to the domain analysis of history and the foundation for the sample analysis presented in the follow-ing sections.

The origin of domain analysis is connected with a pro-posal formulated by two Danish researchers, Birger Hjør-land and Hanne Albrechtsen, in the 1990s (for a short his-tory of a domain analysis see Smiraglia 2015). In a pro-grammatic article published in Journal of the American Society for Information Science, Hjørland and Albrechtsen (1995, 400) advocated studying “the knowledge-domains as thought or discourse communities, which are parts of society’s divi-sion of labor.” Their research program was in contrast to popular then, yet limited approaches based on formal computer-related or cognitive analysis. After this initial step, Hjørland (2002) published the next seminal paper ex-ploring eleven approaches within the domain-analytical



view, such as constructing special classifications and the-sauri, indexing and retrieving specialties, and epistemolog-ical and critical studies. According to Hjørland (2002, 451), the combination of choosing methods and conducting re-search based on them can help “strengthen the identity of IS and strengthen the relationship between research and practice in IS.”

Other authors’ works also contribute significantly to the theory and practice of domain analysis. Hanne Albrechtsen (2015), one of the creators of this new line of research, pre-sented the origins of domain analysis as a new approach to KO and information science in a project on software reuse. Joseph T. Tennis (2003) focused on providing transferable definitions of domains using two analytical devices, namely, “areas of modulation” (the extension of a domain) and “de-grees of specialization” (the intension of a domain). One of the issues addressed by Tennis (2003, 194) was how “the domain is positioned against other domains.” María J. López-Huertas (2015) discussed domain analysis in the light of interdisciplinarity, stressing different requirements for the analysis of disciplines in comparison to interdisciplines. Jenna Hartel (2003) focused on hobby domains as a subject of domain-analytical research. The point of departure for her study was the concept of “serious leisure” coined by Robert A. Stebbins and his taxonomy of hobbies. Hartel provided examples of employing domain knowledge as she, for instance, wrote in the context of analyzing changing forms of recipes, “Literature on the social history of cook-ing would explain how household cooking routines at the turn of the 19th century generated tacit cooking knowledge in children and obviated the need for detailed recipes” (Har-tel 2003, 234). In the context of this paper, it should be un-derscored that there might be real differences between pro-fessional historians’ and hobbyists’ approaches to history, but there has not been such research from the perspective of KO so far. Maurine W. McCourry (2015) presented a model that can be used to assess whether library cataloging codes suit user needs. The author studied a domain of mu-sic, but as she suggested, her methodology could be used to research in other domains as well, e.g., in history.

A vital addition to Hjørland’s work has been made by Richard Smiraglia. He is, among other things, the author of a thorough examination of formal studies employing domain-analytical approaches. The analysis, covering pa-pers published in the years 2003-2014, revealed that the majority of studies used informetric and terminological techniques. Almost no studies focused on the production of guides to reference materials, yet Smiraglia (2015) stressed that they are instead the objects of applied activity in KO. Smiraglia also noted that only five domains had been studied three times as of 2014, namely archives, im-age searching, LGBT, physics, and social media. There were also four papers related to music and twenty-two

dealing with KO. Włodarczyk has also analyzed papers from relevant sources (2015–) such as the journal Knowledge Organization and ISKO proceedings to find additional materi-als using a domain-analytical approach. The analysis re-veals that the state of domain analysis has not changed very much. The domain of KO is still extensively analyzed (e.g., Castanha and Wolfram 2018; Wang 2019). It should be stressed that these kinds of analyses should be seen as fundamental for the future development of KO, but a need exists to embrace other domains, especially within other research communities, like sociology, anthropology, history, and much more. Smiraglia’s work can serve as a point of departure for every project considering domain analysis as its core paradigm. Together with Hjørland’s def-inition (2017a) from the ISKO Encyclopedia of Knowledge Or-ganization, the review prepared by Smiraglia gives insight into the frontier of domain analysis research.

Aside from the contributions presented above, papers on history have been written by KO researchers. Claudio Gnoli (2014) analyzed Marc Bloch’s essay The Historian’s Craft from the perspective of KO. Gnoli (2014, 129) was interested in Bloch’s view on “how historical knowledge can be organized.” His article was divided into four themes: dealing with terminological problems, general rules of the organization of historical knowledge, histori-cal sources, and interdisciplinarity of history. For Gnoli (2014, 134), the most important is the last aspect. He con-cludes: “The very delimitation of a discipline like history involves complex problems. All this should encourage ex-perts of knowledge organization to adopt a cross-medial, interdisciplinary approach, if they really desire to be of help to researchers.” Interdisciplinarity as an important feature of history and how this domain is influenced by other disciplines will be further described in the next sec-tions. It should also be stressed that Bloch’s opinion is only one of many, although formulated by the renowned and influential historian.

Ann M. Graf and Richard P. Smiraglia (2014) carried out a descriptive study of entries related to race and eth-nicity from the bibliography forming a basis for the au-thors of the Encyclopedia of Milwaukee, developed in the his-tory department of the University of Wisconsin-Milwau-kee. They utilized basic bibliometric methods analyzing, among other things, the productivity of authors, types of resources, and title term co-occurrence. Their research showed (Graf and Smiraglia 2014, 120) that “Concept population is shared with the domain at large but without the indicators of productivity that are found in domains representing research fronts.” The paper is a good example of an approach based on quantitative techniques, different from the approach applied by Gnoli (2014). Graf and Smi-raglia (2012) also described the process of the develop-ment of a taxonomy used to organize the content of the



Encyclopedia of Milwaukee. They showed how knowledge of Milwaukee history and practical reasons informed specific decisions made during the activity. Although not explicitly, Małgorzata Pawlak and Karol Sanojca’s (2018) study about the evolution of the internal structure of the Bibliography of Silesia History can also be regarded as KO research in history. The authors presented the changes occurring in the Polish and German bibliographic schemes and indexes.

The temporal aspect of reality has also been studied within KO and related fields. These studies are of both theoretical and practical nature. Jutta Frommeyer (2004) examined the chronological terms and period subdivisions in three subject heading systems, Library of Congress Subject Headings (LCSH), Répertoire d’Autorité Matière Ency-clopédique et Alphabétique Unifié (RAMEAU), and Re-geln für den Schlagwortkatalog, pointing out serious prob-lems connected with these systems. The analysis resulted in a proposal for a new time-retrieval model and search interface. The issue was further explored by Vivien Petras, Ray R. Larson, and Michael Buckland (2006), who pro-posed the Time Period Directories, an infrastructure con-necting named periods, events with dates, and locations. The purpose of it was to allow users to search for temporal information in a more effective way. One of the latest at-tempts to describe periods using linked data is a project entitled PeriodO. It aims to build a period gazetteer that includes different period definitions such as of the Iron Age. The authors (Rabinowitz et al. 2016, 55) hope that due to the improvement of tools associated with a dataset, they “illuminate the evolution of historical disciplines over time.” It appears that with the development of PeriodO, the data can be used by information scientists specializing in KO to better understand the characteristics of temporal information in different disciplines, thereby developing better KOSs.

The researchers have also explored problems related to the cultural aspects of chronological information. Qing Zou and Eun G. Park (2011) presented a proposal for a Chinese Time Ontology, which enables reconciliation with the traditional Chinese approach to time, based on a luni-solar calendar, the titles of different emperors, and their reign periods. They stressed that their model can be em-ployed to other non-Western time scales. The cultural as-pects of temporal information were also explored by Arashanapalai Neelameghan and G. J. Narayana (2013), who presented a detailed study discussing both socio-cul-tural background and time issues concerning KOS.

Studies have been written on historical information sys-tems (HISs) and practical implementations of specific KOSs in history. Javier García-Marco (1994) gave the basic outline of the use of KO in HISs. He described funda-mental issues, like the influence of different theories on terminology, differences in treating historical systems by

heritage (archival) institutions and researchers, and the sketching out of a conceptual system. García-Marco (1994) also observed that a HIS, due to its complexity, is a good testing field for KO. This complexity results both from a long history of a domain and from its interdiscipli-narity. Claire Beghtol (2001) also aimed at addressing a practical problem. She described the methodology em-ployed in the Iter Bibliography, which is a part of the Iter Project, concerned with the Middle Ages and Renaissance, to develop high-quality KO. Further studies investigated the practical uses of different KOSs in history. For in-stance, Branka Purgarić-Kužić (2006) analyzed the practice of classifying materials related to history, according to the Universal Decimal Classification (UDC) in the National and University Library in Zagreb. Apart from the interdis-ciplinarity of history (which is widely acknowledged within the KO community), she stressed, among other issues, the disadvantages of geographical subdivisions for Croatia in UDC. This paper is only one example of numerous studies devoted to different classifications in various cultural set-tings. Although history was a topic of research in library and information science and KO, the analyses are scat-tered, and there is still a lack of theoretical background in the domain of history from the perspective of KO. This research generally did not consider epistemological stances and different discourse communities within a broad do-main of history.

The KO studies on the humanities can be regarded as a source of inspiration for KO research on history. Hjørland (2017a, 441) pointed to Anders Ørom’s (2003) discourse analysis of the domain of art studies as “a model of a do-main-analytic study.” Ørom examined different approaches (paradigms) to art history (traditional paradigms: icono-graphical, stylistic, materialistic, and “new” art history para-digms) and matched them with three levels in the domain of art (art exhibitions, documents, KOSs). The significance of the study lies in the analysis of the impact of different paradigms on the art domain, and consequently, on the the-oretical framework of KOSs in this domain.

Archeology, which is closely related to historical re-search, has also been studied within KO. Teija Oikarinen and Terttu Kortelainen (2013) employed the content de-composition method to research a catalog of archeological artifacts. The study showed problems related to non-uni-formity and omissions in the catalog, which partly resulted from the character of the archeological materials. The au-thors used both qualitative and quantitative analyses to de-scribe the collection description. This approach, based on the combination of different methods, seems to be the most fruitful for domain analysis. It seems to enable infor-mation scientists to look at the analyzed domain from many perspectives complementary to each other. Quanti-tative analysis can provide important basic information



about, among other things, the terms and concepts used, but it does not answer questions concerning the motiva-tions of researchers. It appears to be especially relevant to social science and humanities research (see Hjørland 2002, 439-440). Another study on a domain of archeology is au-thored by Edmund Lee (2017), who presented, among other things, the use of KO in archeological research in the United Kingdom and the impact of KO on new his-torical narratives.

KO studies focused on archival science are of high value for historical research. For instance, Thiago Hen-rique Bragato Barros and João Batista Ernesto de Moraes (2010) carefully analyzed two archival science manuals us-ing discourse analysis to find the differences and similarity within the domain. José Augusto Chaves Guimarães and Natália Bolfarini Tognoli (2015) discussed provenance, a fundamental concept of archival science, as a domain anal-ysis approach. Analyses of archival science from the per-spective of KO seem to be especially important for histo-rians and also for the creators of KOSs in history due to archives being a gateway to primary sources, the essential materials for historical research.

Among the aforementioned studies, Gnoli (2014) and Ørom (2003) are especially interesting in the context of this study because of the objects or methods used. The first is one of few attempts at combining the theory of historiography and KO. The author rightly emphasized the importance of interdisciplinarity in the context of histori-cal research. Ørom is an example of a relatively unusual approach to the humanities in domain analysis based on discourse analysis. Discourse and epistemological analyses can potentially provide the creators of KOSs with addi-tional knowledge to broaden their view of a domain, and thereby lead to the construction of a better KOS (see Hjørland 2002, 439-440).

The KO community can also try to exploit domain knowledge to shape the evolution of KOSs (Smiraglia 2015, 100). Plenty of research studies have been written on historiography and historical methodology, which are highly important for constructing well-suited and useful KOSs in history. Due to the substantial body of literature, only some example sources are mentioned: 1) encyclope-dias and dictionaries regarding historiography, which con-sist of entries for individual historians and regional and topical historiography (e.g., Boyd 1999); 2) books contain-ing an overview of general (e.g., Schneider and Woolf 2011) and national historiography (e.g., Grabski 2006); 3) specific time periods in the history of historiography are also the topic of monographs, particularly about a signifi-cant change in historiography that occurred in the twenti-eth century (e.g., Iggers 2005); 4) numerous individual works and edited books on different fields (e.g., Perks and Thomson 2016), methods (e.g., Gregory et al. 2018), and

specific topics (e.g., Heuman and Burnard 2011); and lastly, 5) scientific journals devoted to the history or theory of historical research, for example, Klio Polska, History and The-ory, Rethinking History, Historical Methods, and Historyka, which can be regarded as useful sources for researchers aiming at analyzing history from the point of view of KO. However, as Georg G. Iggers (2005) underlined, one should be careful about different theories, because some of them differ considerably from the practice of histori-ography. Iggers (2005, 100) wrote that

a number of theorists in France and the United States, mostly coming from literary criticism, such as Roland Barthes, Paul De Man, Hayden White, Jacques Der-rida, and Jean-François Lyotard, frequently identified as postmodernists—a label some of them would vig-orously reject—would call for this surrender and question the distinction between fact and fiction, his-tory and poetry. They viewed history as having no ref-erence to a reality outside of its texts. But as we shall see, practicing historians seldom went so far … Not only did historians continue to work conscientiously and critically with sources, but, … they also adopted methods and findings from the social sciences. Thus, they by no means gave up the conviction that the his-torian must follow rational methods to gain truthful insights into the past.

As a consequence, it is worth noting that, apart from ana-lyzing studies into the theory of history, a more detailed empirical analysis of practical examples of historical writ-ing is needed. This kind of research can utilize, for exam-ple, bibliometric methods like word co-occurrence. The review of the literature presented above shows that no detailed study concerns the broad domain of history from the perspective of KO. Moreover, only a few works apply domain analysis to the humanities. It illustrates that there is a need to conduct more such investigations (cf. Smiraglia 2015). It should be noted, however, that this study is only a preliminary investigation into the domain of history, and more work is needed. 3.0 Wojciech Wrzosek’s theory of historiographical

metaphors This section is devoted to the concept of historiographical metaphor as presented in Wojciech Wrzosek’s (1997) book entitled History, Culture, Metaphor. The rest of this section is based on this author’s description and analysis. The au-thor described a breakthrough in French historiography as a change between the classical and non-classical. Although Wrzosek (1997, 12) focused on a single country’s histori-ography, he stressed that the changes that occurred within



it “reflect certain global trends.” The partial convergence can be seen, for instance, in the case of French and Polish historiography from 1956 through 1989. However, it must be noted that significant background differences existed (Pleskot 2012).

Wrzosek (1997) defined a historiographical metaphor as a fundamental category that forms a basis for historical knowledge, i.e., the works of historians trying to describe and understand the past. It should be stressed that meta-phors are not exclusive to a domain of history but are per-vasive in all domains. As a consequence, scientific meta-phors deserve the KO community’s attention (Hansson 2013). This approach was adopted by Marek Hetmański (2014), who analyzed the role and characteristics of meta-phors in KO.

Wrzosek (1997, 47) emphasized that

The metaphors which are dominant in a culture, in-cluding those on which a certain discipline of schol-arship is based, provide models—so to say—of the field of study.” He later clarified, “Such metaphors, being embedded in culture, become neutral, and (eventually) objective, segments of the commenta-tor’s thinking. In other words, they govern the inter-pretation of phenomena without being controlled by anything. Thus, metaphorical clauses impose a cer-tain understanding of the world, and turn into heu-ristic models of sorts. (47)

The basic information about different historiographical metaphors is presented below. 3.1 Classical historiographical metaphors According to Wrzosek (1997), significant differences exist between classical and non-classical historiography that arose from the change of underlying historiographical metaphors. Classical historiography embraces different schools and approaches, but they all employ such elements as a broadly applied, anthropomorphic view on historical phenomena and traditional historiographical metaphors of development and genesis (with linear time). They are strictly related to each other. The anthropomorphism can be seen not only when the object of the analysis is a life of a single person like a king or a famous hero, but also when traditional historians describe the history of a na-tion, organization, or society. Each subject is treated like an individual who is a direct maker of history, shaping a historical process. The anthropomorphism of classical his-toriography is related to a concept of development and genesis. Wrzosek (1997), citing the work of Robert A. Nisbet, stressed that development is seen by these histori-ans as spontaneous, inborn, continuous, indispensable,

and directional. The metaphor of genesis, important for historiography, can be understood, according to Wrzosek, twofold. First of all, it can be understood as a set of cir-cumstances that are associated with and precede a specific phenomenon (like the genesis of feudalism) and secondly, as “chains of consecutive incidents which are connected by the presumed generation of some by others” (Wrzosek 1997, 46). The concept of genesis, in turn, is related to the concept of time, which is “linear, astronomical time, or-dered in the triad past—present—future” (Wrzosek 1997, 117). 3.2 Non-classical historiographical metaphors Wrzosek (1997) maintained that the change of historio-graphical metaphors resulted in the advent of the French non-classical historiography. The main difference lay in a shift of focus from the “objective” description of related events to the description of human life as a whole. New historians have focused on fields that their traditional counterparts were not interested in, such as everyday life. They started utilizing two main approaches: modernist (quantitative-based social and economic history) and non-modernist (historical anthropology). Wrzosek (1997) di-vided the modernist historiography into moderate and ex-treme. One of the features differentiating them is the ex-istence or the lack of a single determining factor. Wrzosek added that Fernand Braudel’s modernist view does not in-dicate it, but other historians, who are called extreme mod-ernists by him, point to different spheres of social life, e.g., economics or demography, as pivotal points for under-standing historical development and change. The latter ap-proach is related to quantitative history (Wrzosek 1997). The author stressed that modernist historiography refers to new ideas such as Ludwig von Bertalanffy’s systems the-ory, Claude Lévi-Strauss’s theory of structuralism, and Émile Durkheim’s view on sociology. In contrast to tradi-tional historiography, the position of humans in this vari-ant of historiography is significantly different. New histo-rians are more interested in processes and structures than in the lives and activities of individuals, which marks a fun-damental shift in historical research. It marks a departure from the perspective of direct anthropomorphism. The changed metaphors involve a different concept of devel-opment, the modified concept of genesis, “the categories of determinism which are radically different from those of the genetic or cause-and-effect determinism of traditional historiography” (Wrzosek 1997, 99), and the concept of time completely different from linear time. Once again, these metaphors are strictly related to each other. Wrzosek (1997) described these metaphors using the example of Fernand Braudel’s work. According to this view, the devel-opment is seen as non-linear with many possible paths.



The historical world is a complex social system consisting of different subsets such as geographical regions and so-cial groups that are connected by relationships. The system is shaped by different processes, of which the most im-portant are the processes of long duration, almost immu-table, like a relationship between a human and the natural world. These processes determine the identity of the sys-tem. For new historians, a cause-and-effect relationship is not as crucial as for traditional historians who focus on events. They use a different intra-structural concept of genesis, where links between different substructures and changes of substructures are the most important deter-mining factors. The cause-and-effect determinism has been replaced by functional determinism. A new kind of determinism also influences the concept of time. As Wrzosek (1997, 117-118) stressed,

modernist historians are no longer able to assign pre-cise dates to patterns of economic evolution, booms and slumps in the market or instances of cultural dif-fusion; they may only indicate the duration of such developments and their temporal correlation with other processes. This is because it is impossible to assign a precise date to such occurrences as Protes-tantism, the baroque or the influence of Islam.

Wrzosek (1997) wrote that the advent of non-modernist historiography has originated from the objection to the structuralist and functionalist metaphors of modernist his-toriography based on economics and sociology. The crea-tors of this new line of research have turned to anthropol-ogy as a focal point. The central concept has become the culture and people as actors in different cultural settings. According to Wrzosek (1997, 141):

The genre’s luminaries, along with foreign (i.e., non-French) scholars who follow in their footsteps, try to preserve such components of the ideology of the An-nales school as would allow the maintaining of tradi-tional historiography’s humanistic value: the tenet that the human being is the focus of history. They want to sustain the human being’s prominent position. Never-theless, the human being is now deemed not the agents of events, but the creator of and participant in culture, and the vehicle of civilization. By means of examining the individual or an anonymous group of individuals, we may fathom the universal and supra-individual rules that inflexibly govern a certain society or epoch.

The important line of research has become investigation on mentalité (mentality), understood by Georges Duby in his article Histoire des mentalités from 1961 (as cited in

Wrzosek 1997, 136) as “a system of images and concepts that produce different harmonies in the various social groups and strata making up society.”

Wrzosek (1997) explored and described two distinct ap-proaches to historiography, classical and non-classical, that emphasize different aspects of historical existence. He noted (1997, 124-125),

The two images of the human being, the traditional one and that depicted by non-event historiography, are mutually incompatible. It seems impossible to develop a coherent historical narrative that would use both approaches—the process-centered and event-based. In order to do this after all, one would have to establish explanatory links between process-centered and event-based reality, inter alia, by proving that functional determinism may be reduced to cause-and-effect brand—or vice versa; if this is unfea-sible, then both spheres of discourse remain incom-mensurable with respect to explanation.

This problem was exemplified by Wrzosek (1997) by Fer-nand Braudel’s (1995) work entitled The Mediterranean and the Mediterranean World in the Age of Philip II, which failed to achieve this goal.

The historiographical metaphors can be used as a high-level backdrop for analyzing KOSs in history and indexing historical works. The next section contains a review of sample universal indexing tools used in Polish libraries from the perspective of historiographical metaphors, and consequently traditional and “new” history. These meta-phors help to uncover the epistemological backbone of these systems and identify possible improvements and changes.

4.0 KABA subject headings and the National

Library of Poland Descriptors from the perspective of different historiographical traditions and metaphors

The section presents two Polish universal KOSs through the lens of classical and non-classical historiography and Wrzosek’s historiographical metaphors. First, both sys-tems are briefly introduced. Secondly, the structure of KABA and NLPD and indexing practices are presented in relation to classical historiographical metaphors. Lastly, these systems are analyzed from the standpoint of non-classical metaphors and interdisciplinarity as a basis of non-classical historiography. Examples of the structure of KOSs and book descriptions are representative; that is, they have been selected to illustrate standard practices fol-lowed by developers and indexers.



4.1 Introduction to KABA subject headings and the National Library of Poland Descriptors

The two Polish universal systems analyzed here differ sig-nificantly from each other, as their history, structure, and indexing rules are markedly different. KABA, managed by the NUKAT Center, is used by the network of Polish ac-ademic and scientific libraries. It originated from work car-ried out in the early 1990s aiming to automate library pro-cesses at the Warsaw University Library. After careful con-sideration, it was decided that the basis for the creation of a subject access tool would be a French subject heading system, RAMEAU, with Library of Congress Subject Headings (LCSH) and Répertoire des vedettes matière de l’Univer-sité Laval as secondary sources. The French subject head-ings were a source of relational structure and rules of ap-plication of specific headings and subdivisions for the Polish thesaurus (Kotalska 2002). Barbara Kotalska (2002, 154), an author of the article presenting the cooperation between KABA and RAMEAU teams, cautioned that “RAMEAU headings are subjected to a naturalization pro-cess.” However, this translation process sometimes re-sulted in the use of headings and subdivisions inappropri-ate for Polish historiography. For instance, a chronological subdivision “1870-1914” copied from RAMEAU is used in KABA to index books about Polish history. This period is especially significant for France and Germany, but not for Poland, because the starting date is connected with the outbreak of the Franco-German War. However, with time, KABA has gained more independence; i.e., more and more decisions have been made independently, yet older head-ings and subdivisions have not been corrected.

The team responsible for the development of KABA consists of subject specialists. However, each person usu-ally manages vocabulary for more than one discipline. For example, Iwona Ruść deals with economics, geography, psychology, pedagogy, and sociology. Paweł Rygiel is re-sponsible not only for history, but also for law, religious, and art studies (NUKAT 2018). It seems impossible that one person could have substantial knowledge of all these domains. For instance, Rygiel is a specialist in archeology. Consequently, the question arises whether the KABA vo-cabulary for different domains is developed at the same level of granularity.

Although there were some attempts to use KABA in a novel way (e.g., Mazurek et al. 2014), it has remained a tra-ditional, pre-coordinated system based on headings and subdivisions used mainly in Polish academic libraries. The team responsible for the development of the KABA sys-tem has made some improvements, e.g., it has introduced new genre/form headings in line with changes introduced in the Library of Congress. It should be noted, however, that the accepted rules are complex in comparison to the

Library of Congress Genre/Form Terms for Library and Archival Materials (Włodarczyk 2014).

NLPD, the system developed in the National Library of Poland, was introduced at the beginning of 2017. It has been recently presented by Joanna Cieloch-Niewiadomska (2019), so in this article, only some fundamental rules of this system are presented. Apart from the article men-tioned above, information about the project is also availa-ble at the special website maintained by the National Li-brary of Poland (http://przepisy.bn.org.pl/). The descrip-tion provided below is based on these documents. The main source of the vocabulary, at least in the first stage of the project, is the previous system called the National Li-brary of Poland Subject Headings (NLPSH), which uti-lized the division into headings and subdivisions. The NLPSH thesaurus is now being transformed according to the rules of NLPD, and after the completion of this pro-cess, it will serve as a basis for retrieval and indexing. The analysis of the NLPD structure presented in the article is performed using examples from this unfinished tool, but ready-to-use descriptors have been selected.

In contrast to NLPHS, NLPD is a post-coordinated system based to a large extent on common usage. In com-parison to KABA, the vocabulary is characterized by a much more specific level of granularity; i.e., the creators have tried to separate individual pieces of information as much as possible. NLPD is formed by a set of independ-ent, domain-specific thesauri, but connected and pre-sented in one interface. This structure of separate domain-specific thesauri seems to ease the issue of domain knowledge incorporation but may restrict interdisciplinary development of the system.

The NLPD vocabulary is divided into twelve facets: “form and type,” “genre,” “audience,” “cultural area,” “subject,” “subject: person,” “subject: corporate body,” “subject: event,” “subject: place,” “subject: work,” “sub-ject: time,” and “discipline and approach.” The structure described above is not fully employed due to practical lim-itations of the software used by the National Library of Poland, where only nine facets are available, and some de-scriptors are wrongly assigned to facets. For instance, genre descriptors are presented in the general “subject” facet.

The development of NLPD’s domain-specific vocabu-lary is addressed differently from the solution adopted at the NUKAT Center. During the NLPD project starting in 2015, a number of subject librarians, e.g., in history, eco-nomics, and ethnology and anthropology, have been em-ployed for, among other things, “managing subject the-sauri, and designing new descriptors according to the de-mands of the publications within the discipline and this discipline’s development” (Cieloch-Niewiadomska 2019, 46). Cieloch-Niewiadomska (2019) provided examples of



vocabulary enhancement resulting from the employment of specialists. 4.2 KABA subject headings and the National

Library of Poland Descriptors from the perspective of classical historiography and its historiographical metaphors

4.2.1 The structure of KOSs The classical metaphors of development and genesis can be tracked in the structure of both KOSs. This type of meta-phor is especially prevalent in the context of event-related political historiography. Thesauri may show the genetic se-quence of events and the linearity of development. To ana-lyze KOSs from the perspective of classical historiograph-ical metaphors, they should contain well-developed vocabu-lary related to different kinds of events. Both KABA and NLPD are in line with these requirements in this respect. According to classical metaphors, these events can be easily presented on a timeline as points in time because they are more or less precisely dated and linked by a cause-and-effect relation. As an example for further analysis, the shortened, translated versions of records representing the First World War are presented and analyzed below.

NLPD World War I (1914-1918) RT Paris Peace Conference (1919-1920) RT Assassination at Sarajevo (1914) NT Battle of Gorlice (1915) NT Battle of Konary (1915) NT Battle of Łowczówek (1914) Polish Legions (1914-1917) RT Battle of Konary (1915) RT Battle of Łowczówek (1914) KABA World War, 1914-1918 RT World history—20th century World War, 1914-1918—Campaigns and battles—Po-

land NT Gorlice, Battle of (1915) NT Konary, Battle of (1915) NT Łowczówek, Battle of (1914) World War, 1914-1918—Causes RT Sarajevo (Bosnia and Herzegovina)—1914 (Assas-

sination)

In the light of traditional historiographical metaphors, it is not enough to code the exact date of the event, e.g., the day and time of a battle. The genetic relationships between the events need to be explicitly shown. The need for im-plementation of such causal relations in information sys-tems has been already identified by different researchers (e.g., Petras et al. 2006). In the records presented above, several traces of this way of thinking can be found. First of all, it can be seen in the fragment related to the assassi-nation of Archduke Franz Ferdinand of Austria. The de-scriptor in NLPD, which represents this event, is directly related to the descriptor “World War I (1914-1918),” which reflects the genetic relationships between the assassination and the outbreak of the First World War. In contrast, the creators of KABA have added another level of the divi-sion with the subdivision “Causes,” whose name directly refers to a metaphor, yet it is the only event mentioned in this record. Additionally, the term “World War I (1914-1918)” in NLPD is related to a descriptor representing the Paris Peace Conference. This way of thinking is not in-cluded in KABA since headings for meeting names are not connected with topical terms. The examples provided above are also in line with a linear time metaphor.

However, the two systems do not include other associ-ative relationships that can be established through the point of view of classical historiographical metaphors. The set of narrower terms for the First World War in NLPD is formed by the names of different battles, whereas in KABA, once again, in compliance with the grammar of the system, there is one more hierarchical level established by the addition of subdivisions “cam-paigns and battles,” supplemented by the names of the countries or regions. When one looks at these examples from the perspective of classical metaphors, new associa-tive relationships could be established between related events. It should be underscored that several choices al-ways exist. For example, battles form different lines of successive stages. In the battle of Łowczówek (22-25 De-cember 1914), the First Brigade of the Polish Legions, which fought on the Austro-Hungarian side, faced the Russian forces. After the battle, which did not change the strategic situation considerably, the next more significant clashes at this part of the Eastern front line broke out dur-ing Gorlice-Tarnów Offensive of Austro-Hungarian troops. It was the logical aftermath of the battle of Łowczówek, and consequently, an associative relationship could be established. The next battle fought by the First Brigade was the battle of Konary (Klimecki 1993), so it could also be related to the descriptor representing the bat-tle of Łowczówek. These are only two examples of possi-ble choices. In comparison to KABA, the option for the creation of such relationships in NLPD is facilitated due to the establishment of the relationships between the de-



scriptor for the Polish Legions and descriptors represent-ing different battles in which Polish soldiers fought.

It should be noted, however, that there are serious problems associated with the classical metaphor of devel-opment. Wrzosek (1997) contended that this metaphor has become intertwined with other metaphors, e.g., of pro-gress and evolution. For the history of a concept of state, for instance, it means that there are initial, intermediate, and final stages of development. The last one can be dif-ferently conceptualized, for instance, as Western democ-racy or a communist state but always as a highly desirable and final stage. Contemporary research and practice show that this view is too simplistic; hence, it cannot be (and is unadvisable to be) directly included in the KOS structure. It shows the limitations of including traditional metaphors when other modernist and anti-modernist approaches are vital and accepted within the research community. 4.2.2 Indexing practices Indexing can also be analyzed from the point of view of classical historiographical metaphors. For instance, it is possible to do so with a traditional metaphor of genesis. Once again, it is primarily possible in the case of some po-litical and military history works, i.e., the books and articles that emphasize and analyze the roots of political events. An example of such an approach is Christopher Clark’s (2013) book entitled Sleepwalkers, whose aim is to explain the decisions made by politicians that led to the outbreak of the First World War. As Clark (2013, xxix) wrote,

The story this book tells is, by contrast, saturated with agency. The key decision-makers—kings, emperors, foreign ministers, ambassadors, military commanders and a host of lesser officials—walked towards danger in watchful, calculated steps. The outbreak of war was the culmination of chains of decisions made by polit-ical actors with conscious objectives, who were capa-ble of a degree of self-reflection, acknowledged a range of options and formed the best judgements they could on the basis of the best information they had to hand. Nationalism, armaments, alliances and finance were all part of the story, but they can be made to carry real explanatory weight only if they can be seen to have shaped the decisions that—in combi-nation—made war break out.

The descriptions of the 2017 Polish translation of the book in both Polish KOSs are provided below. Both sys-tems employ MARC 21 Format for Bibliographic Data. MARC tags are given only to identify data elements used (cf. Library of Congress 2017), to understand the exam-ples better.

KABA 650 World War, 1914-1918—Causes 650 World War, 1914-1918—Diplomatic history 651 Europe—Politics and government—1871-1918 NLPD 648 1801-1900 648 1901-2000 648 1901-1914 648 1914-1918 650 World War (1914-1918) 650 World politics 651 Europe

As described in Section 3.1, the establishment of genetic relationships between events is one of the essential foun-dations of historical research employing traditional meta-phors. There is no one solution used in KABA to show the genesis of events at the level of indexing. The first solu-tion, exemplified by the description given above, utilized two groups of headings: the name of an event with the subdivision “causes” and strings representing facts or do-main where the genesis of this event was sought. The sec-ond solution is used when a book or article is devoted to many different factors contributing to the event described. In such instances, only the name of an event supplied by the subdivision “causes” is utilized. The question arising from the second solution is connected with the granularity of a description: is it possible to take into account more specific factors? It could be useful, especially in the context of academic and research libraries, but the level of granu-larity should be set independently by each institution.

The NLPD indexing rules do not allow indexers to in-dicate the genesis of events explicitly. The difficulties re-sult partly from the characteristics of post-coordinated systems. There is no direct indication of cause and effect descriptors, which are listed in a description on the same level. However, admittedly, analyzed subject indexing prac-tices have not been applied consistently. The element of causation is not included in numerous cases that it could be in both KABA and NLPD. In the NUKAT Center, it seems to be connected with the practice of a central cata-log where librarians from different institutions produce records, including subject indexing.



4.3 KABA subject headings and the National Library of Poland Descriptors from the perspective of non-classical historiography and its historiographical metaphors

4.3.1 The structure of KOSs Many historical works written in the twentieth century ex-emplify the employment of non-traditional approaches that one can also try to include in the process of creating thesauri. These approaches are connected, according to Wrzosek (1997), with the influence of other domains like sociology, economics, ethnology, and anthropology; hence their terminology should be incorporated into KOSs at an appropriate level of granularity. It seems that at the stage of vocabulary development, taking into account non-clas-sical metaphors is, to some extent, equal to the analysis of interdisciplinary issues connected with new lines of re-search. Such a thesaurus needs to include many interdisci-plinary terms; however, it should be underscored that the creators of thesauri in the discipline of history need to an-alyze primarily terms developed within this domain. The analysis shows that there is a lack of some important terms in both Polish KOSs. For instance, in NLPD, one cannot find the term “microhistory,” which is a significant sub-field of historical anthropology.

The systems built according to non-classical ap-proaches should include the terms for different disciplines, subdisciplines, fields, subfields, but what is equally im-portant is the appropriate relationships pointing to the im-pact and interrelations between them. This condition is better satisfied by the NLPD system, which has a better-developed network, consistent with contemporary thesau-rus construction standards and Polish historical research. In KABA, for instance, the descriptor “social history” is not related to the term “history” but only to “sociology;” similarly, the descriptor “economic history” is only con-nected to the term “economics.” These are examples of the general situation that needs to be improved.

Another aspect connected with the advent of new his-toriography that needs to be emphasized is related to broadening the scope of research methods and techniques. Modernist historians, due to the interest in a wide range of topics, have accommodated methodological influences from different disciplines, and within them from different scientific theories. Both systems include descriptors repre-senting them, but the main difference lies in the method of categorizing terms. In KABA some terms, e.g., “statis-tical methods,” are represented as subdivisions, which are not included in the network of relationships, because there are no relations between headings and subdivisions. This lack is a permanent, well-known feature of such systems that impedes their browsing functionality. In contrast to

KABA, NLPD has better-developed relationships be-tween the terms denoting methods and techniques. How-ever, it seems that there is a need to supplement the two systems with more terms for methods used in contempo-rary historical research.

Another issue, which seems to be primarily connected with the complexity of new historiography, concerns the types of relationships that should be included in KOSs. This problem, but from the general point of view, was discussed by Hjørland (2016) in an article about the future of tradi-tional thesauri. Hjørland (2016, 152) noted that “Different kinds of relations have different importance in different do-mains . . . Because of this, and because the semantic rela-tions themselves may be relative to different queries, it is problematic to consider a thesaurus as a uniform standard for all fields as opposed to a domain-specific semantic tool.” Hjørland (2016) argued, discussing the differences between thesauri and ontologies, that semantic relations should not be restricted to three basic types (BT/NT, RT, USE/UF) at any cost, because there is no research indicating the atem-poral and domain-independent value of such an approach. Both KABA and NLPD only utilize these basic types of re-lationships. It seems that the establishment of more specific relationships in KOSs in history could facilitate the descrip-tion of a complex landscape of contemporary historiog-raphy based on different metaphors. For instance, a special type of relationship could be used to show the connection between influencing and influenced substructures. How-ever, this issue has to be further investigated in the context of specific history works.

Interdisciplinarity that influences both terms and rela-tionships among them is closely related to new historio-graphical metaphors. One of them is a new metaphor of genesis. Instead of the line of events where one generates another linearly, new historians have proposed more com-plex solutions of intra-structural genesis. They are not ho-mogenous; that is, there are numerous variants of this gen-eral approach. Some of them identify a specific aspect of reality as a causative factor, and some build a more multifac-eted image of historical reality (Wrzosek 1997). In the con-text of the former approach, a KOS creator might relate different fields to one selected as the most important one. For instance, the creation of a system from the perspective of the theory that the historical reality is founded on eco-nomics entails the extension of terms and relationships re-lated to this domain, and moreover, that is fundamental in the light of the theory, the subordination of different as-pects to the economy. The subordination does not neces-sarily need to be understood as the creation of hierarchical relationships but rather as the direct linkage between de-scriptors representing different aspects of reality with that representing economic phenomena. However, the richness of modern historical research in terms of methodological



and theoretical pluralism indicates that the structure of the majority of KOSs cannot be founded on a single approach to historiography such as that assuming the precedence of the economy over other domains. As a consequence, the task of establishing relationships in compliance with this in-tricate image is much more difficult in comparison to a sys-tem based on a single approach. Every aspect of reality, like a structure or process, can be potentially related to others. It seems that the only solution for the creators of KOSs is to analyze the existing approaches carefully and decide which are the most representative and beneficial for their group of users. It can be achieved by a thorough examination, not only of the history domain, but also other related disciplines and fields. During this activity, there is a need to include sub-ject specialists to a much greater extent than before. As was presented in Section 4.1, this aspect is addressed better in the development of NLPD. 4.3.2 Indexing practices An important issue, worth considering from the perspec-tive of non-classical historiography and its metaphors, concerns how history works are indexed in both systems. One of the approaches used in non-classical historiog-raphy is historical anthropology. An example of such an approach is a book entitled A History of the Body in the Mid-dle Ages by Jacques Le Goff and Nicolas Truong (2018), which presents the two-fold approach of Medieval people to the body as both positive and negative. Different as-pects of the main object, such as gender, work, crying, dreams, age, death, eating, beauty, and social metaphors re-lated to the body, are described and analyzed. Le Goff and Truong have tried to approach the topic, from, among oth-ers, sociological and anthropological perspectives, absent in traditional history, and consequently not embraced by classical metaphors. Although they described the situation from the fifth to fifteenth centuries, with some remarks on the sixteenth and seventeenth centuries, the book cannot be considered a systematic and thorough description of the issues mentioned above. It is instead a historical essay that can be regarded as an example of an historical anthro-pological analysis, including the history of the mentality of medieval people. The results of the subject indexing of the work are presented below.

KABA 650 Human body—Religious aspects—Middle ages 650 Sociology of the body—Middle ages NLPD 045 1001-1501

648 1101-1200 648 1201-1300 648 1301-1400 648 1401-1500 648 1501-1600 650 Human body 650 Philosophical anthropology 650 Social anthropology 650 Sociology of the body 650 Middle Ages 651 Europe 655 Essay 658 Ethnology and cultural anthropology 658 History

Apart from the lack of precision (the book is mainly about Western Europe), librarians from both institutions correctly used the name of the leading topical element: “human body.” The interesting aspect of these descriptions lies in the use of descriptors representing different disciplines and their branches. The use of the descriptor “sociology of the body” in KABA’s description results from the equivalence relationship with “human body—social aspects” and “hu-man body—sociology.” By contrast, the description made using NLPD involves the disciplinary approach of the book (“ethnology and cultural anthropology,” “history”) de-scribed explicitly in the separate 658 MARC fields (see Li-brary of Congress 2017). The creators of the system have allowed librarians to use only thirty-four selected descriptors to show this feature, and consequently, some disciplines and subdisciplines, although also used by researchers, cannot be employed this way. The NLPD description of A History of the Body in the Middle Ages also includes the descriptors for the sociology of the body, philosophical, and social (cul-tural) anthropology that cannot be used to show the disci-plinary point of view adopted by the authors. However, these descriptors do not show the subject of the book, but rather the authors’ perspective on the subject. The essay is undoubtedly not about philosophical or social anthropol-ogy. Moreover, although the NLPD contains the descriptor “historical anthropology,” it was not used to describe the book. It shows that the NLPD system should be developed further to accommodate different disciplines’ perspectives.

As discussed, non-classical historiography also employs new research methods from a range of social sciences disci-plines. However, currently, neither system takes into account this property of historical writing. The decision about the inclusion of such information could be based, e.g., on user studies similar to the one conducted by McCourry (2015).

Another important aspect is the method of indexing temporal information in both systems. Time is metapho-rized differently in non-classical historiography, which also has some implications for KO. It has not been seen as



merely linear and teleological. Wrzosek (1997) drew the connection between functional determinism and a new concept of time. He showed this aspect of modernist his-toriography through the example of Fernand Braudel’s work. According to the French historian, different aspects of reality evolve in different rhythms. Some of them change more slowly than others. One of the examples of slow development is, according to Braudel, material cul-ture from the fifteenth to the eighteenth century (see Wrzosek 1997). The subject descriptions of the first vol-ume of Braudel’s (2019) Civilization and Capitalism, 15th-18th Century, which focus on this aspect of reality, are pre-sented below.

KABA 650 Economic history—15th century 650 Economic history—1500-1800 650 Material culture—15th century 650 Material culture—1500-1800 NLPD 648 1401-1500 648 1501-1600 648 1601-1700 648 1701-1800 650 Economic history 650 Capitalism 650 Material culture 651 World 655 Monograph 658 Economy, economics, finance 658 History

The use of different chronological subdivisions or de-scriptors suggests the lack of continuity, which does not comply with Braudel’s view. This issue is connected with what is called by Hjørland (2017b, 59-60) “the epistemo-logical view” of subjects, which asserts “that different ‘par-adigms’ entail different subject representations.” If one agrees that the “long duration” concept is cognitively im-portant, and it seems to be such, the creators of KOSs of history should take it into account while formulating in-dexing rules. Braudel’s book is not merely a collection of simple facts and events that occurred around the world from the fifteenth to eighteenth centuries. Braudel looked instead at the durability of the world in longer period. In the context of such works, the idea of using the names of the periods mentioned by Petras, Larson, and Buckland (2006), which suggest durability, seems to be in line with the “long duration” underlined by Braudel. It would also be possible to add additional descriptors that could em-

phasize whether an author of a historical book or article looked at subject from the perspective of long term pro-cesses, e.g., with the descriptor “long duration.”

In this section, KABA and NLPD systems have been analyzed through the lens of different historiographical traditions and metaphors. Both structures of KOSs and indexing practices have been examined on selected exam-ples attempting to show that the knowledge of different traditions and metaphors can illuminate the development of such systems. It seems that due to this knowledge, KOSs creators can make better, informed decisions. 5.0 Summary This article has presented a general review of literature, which is crucial from the perspective of the creation of KOSs in history. Moreover, it has provided examples of epistemological analyses of KOSs’ structures and indexing practices of history books from the point of view of a single historiographer’s framework. There are other theo-retical standpoints in history that should be analyzed from the perspective of their applicability to the development of KOSs. It seems that even such a general epistemological theory as presented in this paper can be regarded as a val-uable point of departure for analyzing and possibly im-proving KOSs in history. KO researchers and practitioners need to be aware of this theoretical background that can bring changes both to thesaurus structure and indexing practices. However, the designers of systems, both in their structure and indexing rules, will inevitably face the prob-lem of deciding which specific elements can be included in a system; that is, they need to compromise on the ques-tion of domain knowledge. As Wrzosek (1997) stressed, two distinct approaches to historical reality represented by classical and non-classical historiography are difficult, or even impossible, to combine in historical writing. It seems that there are two main possible solutions to address the challenge faced by KOS designers. The first solution is to build a system based on only one group of interrelated metaphors, but it would result in the exclusion of other views (cf. the example of BKK provided by Ørom 2003, 142). It would be potentially possible to do so only when information resources indexed with this system would be limited to representing this one group of metaphors. The second solution is to attempt to take into account different groups of metaphors and to try to include them into one system. This solution is associated with a similar issue that was faced by Ørom (2003) during the analysis of the visual art domain. Ørom (2003, 142) wrote,

Because the Art & Architecture Thesaurus is a more “open” and more expanded work of “bricolage” than universal classification systems, it is easier to in-



tegrate new aspects of art studies in the facet struc-ture. At a theoretical level however, the eclecticism and the “additive” conception of conceptual rela-tions mean that the Art & Architecture Thesaurus has a problematic epistemological foundation.

In the case of KOSs in history, different facets could be organized according to different traditions and metaphors. Indexers could also possibly apply knowledge about clas-sical and non-classical historiography to improve indexing practices. Some examples were provided in Sections 4.2.2 and 4.3.2, but more work is needed.

An additional problem can result from a possible lack of domain knowledge. Non-specialists creating domain vocab-ulary and indexing history books and articles may not notice the need and possibility of different choices, as primary and secondary school curricula are based on traditional meta-phors for the most part. Therefore, the employment of spe-cialists is advised. This is accommodated in both Polish sys-tems, yet, as it was presented in Section 4.1, the developers employed at the NUKAT Center also need to deal with the vocabularies outside of their areas of expertise. Similarly, in-dexing books at NUKAT is often done by librarians who are not specialists in a specific domain.

Although the knowledge about historiographical meta-phors given by Wrzosek can be helpful for the analysis and creation of KOSs, it seems that their broad character can provide the creators only with some general guidelines. His-torical research is multidimensional, which is why the gen-eral remarks presented in this article need to be supple-mented with in-depth theoretical and empirical analyses of history; that is, there is a need to analyze thoroughly differ-ent branches of history, fields, and subfields. According to Tennis (2003), it is equal to diminish the extension of a do-main and to increase its intension. The basis for such an in-vestigation can be books and papers on theory and history of historiography, some of which have been reviewed in Section 2.0 of the paper. Moreover, historical works and user studies can be conducted to supplement analysis. It ap-pears that such a line of research can provide the basis for creating user-friendly and domain-grounded KOSs in his-tory. Note 1. Many other disciplines use historical methods and ana-

lyze their subjects chronologically, e.g., musicology com-prises historical musicology that focuses on the history of musical genres and life of composers, but this domain is distinct from the discipline of history due to its differ-ent aims, traditions, and primary objects of inquiry. As Richard T. Vann (2018) wrote about the history of the arts: “Despite essential differences, these forms of histo-

riography have some common features. One is that they are almost invariably produced outside history depart-ments and faculties. For this reason they have tended to be regarded as somewhat exotic specialties.” Cf. Gnoli (2014, 133-14) who discussed boundaries of history.

References Albrechtsen, Hanne. 2015. “This is Not Domain Analy-

sis.” Knowledge Organization 42: 557-61. Barros, Thiago Henrique Bragato and João Batista Ernesto

de Moraes. 2010. “From Archives to Archival Science: Elements for a Discursive Construction.” In Paradigms and Conceptual Systems in Knowledge Organization: Proceedings of the Eleventh International ISKO Conference 23-26 February 2010 Rome, Italy, ed. Claudio Gnoli and Fulvio Maz-zocchi. Advances in Knowledge Organization 12. Würz-burg: Ergon, 398-404.

Beghtol, Clare. 2001. “Knowledge Representation and Or-ganization in the Iter Project: A Web-Based Digital Li-brary for Scholars of the Middle Ages and Renaissance (http://iter.utoronto.ca).” Knowledge Organization 28: 170-79.

Boyd, Kelly, ed. 1999. Encyclopedia of Historians and Historical Writing. 2 vols. London: Fitzroy Dearborn.

Braudel, Fernand. 1995. The Mediterranean and the Mediterra-nean World in the Age of Philip II, trans. Siân Reynolds. 2 vols. Berkeley: University of California Press. Trans. of Méditerranée et le monde méditerranéen à l'époque de Philippe II.

Braudel, Fernand. 2019. Struktury codzienności: możliwe i nie-możliwe, trans. Maria Ochab and Piotr Graff. Warszawa: Państwowy Instytut Wydawniczy. Trans. of Civilisation matérielle, économie et capitalisme.

Castanha, Renata Cristina Gutierres and Dietmar Wolf-ram. 2018. “The Domain of Knowledge Organization: A Bibliometric Analysis of Prolific Authors and Their Intellectual Space.” Knowledge Organization 45: 13-22.

Chaves Guimarães, José Augusto and Natália Bolfarini Tognoli. 2015. “Provenance as a Domain-Analysis Ap-proach in Archival Knowledge Organization.” Knowledge Organization. 42: 562-69.

Cieloch-Niewiadomska, Joanna. 2019. “Introducing the National Library of Poland Descriptors to the Polish National Bibliography.” Cataloging & Classification Quar-terly 57, no. 1: 37-58. doi: 10.1080/01639374.2019. 1573774

Clark, Christopher. 2013. The Sleepwalkers: How Europe Went to War in 1914. New York: HarpersColllins.

Finney, Patrick. 2005. “What is International History?” In Palgrave Advances in International History, ed. Patrick Fin-ney. London: Palgrave Macmillan, 1-35.



Frommeyer, Jutta. 2004. “Chronological Terms and Period Subdivisions in LCSH, RAMEAU and RSWK: Devel-opment of an Integrative Model for Time Retrieval across Various Online Catalogs.” Library Resources & Technical Services 48, no. 3, 199-212.

García-Marco, Javier. 1994. “Knowledge Organisation in Historical Information Systems.” In Knowledge Organiza-tion and Quality Management: Proceedings of the Third inter-national ISKO Conference 20-24 June 1994 Copenhagen, Den-mark, ed. Hanne Albrechtsen and Susanne Oernager. Advances in Knowledge Organization 4. Würzburg: Er-gon, 81-90.

Gnoli, Claudio. 2014. “Boundaries and Overlaps of Disci-plines in Bloch's Methodology of Historical Know- ledge.” In Knowledge Organization in the 21st Century: Between Historical Patterns and Future Prospects: Proceedings of the Thir-teenth International ISKO Conference 19-22 May 2014 Kraków, Poland, ed. by Wiesław Babik. Advances in Knowledge Organization 14. Würzburg: Ergon, 129-35.

Grabski, Andrzej Feliks. 2006. Zarys historii historiografii polskiej. Poznań: Wydawnictwo Poznańskie.

Graf, Ann and Richard P. Smiraglia. 2012. “Cultural Cura-tion as Classification: The Evolution of the Bibliography and Taxonomy for the Encyclopedia of Milwaukee.” In Information in a Local and Global Context: Proceedings of the 40th Annual Conference of the Canadian Association for Infor-mation Science, Waterloo, Ontario, May 31-June 2, 2012, ed. by Anabel Quan-Haase, Victoria L. Rubin and Debbie Chaves. https://journals.library.ualberta.ca/ojs.cais-acsi. ca/index.php/cais-asci/article/view/640/590

Graf, Ann M. and Richard P. Smiraglia. 2014. “Race & Eth-nicity in the Encyclopedia of Milwaukee: A Case Study in the Use of Domain Analysis.” In Knowledge Organization in the 21st Century: Between Historical Patterns and Future Pro-spects: Proceedings of the Thirteenth International ISKO Confer-ence 19-22 May 2014 Kraków, Poland, ed. by Wiesław Babik. Advances in Knowledge Organization 14. Würzburg: Er-gon, 114-20.

Gregory, Ian, Don DeBats and Don Lafreniere, ed. 2018. The Routledge Companion to Spatial History. London: Routledge.

Hansson, Joacim. 2013. “The Materiality of Knowledge Organization: Epistemology, Metaphors and Society.” Knowledge Organization 40: 384-391.

Herodotus. 1996. Histories, trans. George Rawlinson. Ware: Wordsworth Editions.

Hetmański, Marek. 2014. “The Actual Role of Metaphors in Knowledge Organization.” In Knowledge Organization in the 21st Century: Between Historical Patterns and Future Pro-spects: Proceedings of the Thirteenth International ISKO Confer-ence 19-22 May 2014 Kraków, Poland, ed. by Wiesław Babik. Advances in Knowledge Organization 14. Würzburg: Er-gon, 73-79.

Heuman, Gad and Trevor Burnard, ed. 2011. The Routledge History of Slavery. London: Routledge.

Hjørland, Birger. 2002. “Domain Analysis in Information Science: Eleven Approaches —Traditional as Well as In-novative.” Journal of Documentation 58: 422-62. doi: 10. 1108/00220410210431136

Hjørland, Birger. 2016. “Does the Traditional Thesaurus Have a Place in Modern Information Retrieval?” Know- ledge Organization 43: 145-59.

Hjørland, Birger. 2017a. “Domain analysis.” Knowledge Or-ganization 44: 436-64.

Hjørland, Birger. 2017b. “Subject (of Documents).” Knowledge Organization 44: 55-64.

Hjørland, Birger and Hanne Albrechtsen. 1995. “Toward a New Horizon in Information Science: Domain‐Analy-sis.” Journal of the American Society for Information Science 46: 400-25. doi: 10.1002/(SICI)1097-4571(199507)46:6 <400::AID-ASI2>3.0.CO;2-Y

Iggers, Georg G. 2005. Historiography in the Twentieth Century: From Scientific Objectivity to the Postmodern Challenge. Mid-dletown: Wesleyan University Press.

Klimecki, Michał. 1993. Łowczówek 1914. Warszawa: Wydawnictwo Bellona.

Kotalska, Barbara. 2002. “The RAMEAU/KABA Network: An Example of Multi-Lingual Cooperation.” Slavic & East European Information Resources 3, no. 2-3, 149-56. doi: 10.1300/J167v03n02_15

Le Goff, Jacques and Nicolas Truong. 2018. Historia ciała w średniowieczu, trans. Ireneusz Kania. Warszawa: Wydawnictwo Aletheia. Trans. of Histoire du corps au Moyen Âge.

Lee, Edmund. 2017. “Knowledge was their Treasure: Ap-plying KO Approaches to Archaeological Research.” Knowledge Organization 44: 644-55.

Library of Congress. 2017. “6XX - Subject Access Fields-General Information.” https://www.loc.gov/marc/ bibliographic/bd6xx.html

López-Huertas, María J. 2015. “Domain Analysis for In-terdisciplinary Knowledge Domains.” Knowledge Organi-zation 42: 570-80.

Mazurek, Cezary, Krzysztof Sielski, Justyna Walkowska and Marcin Werla. 2014. “KABA Subject Heading Language as the Main Resource Subject Organization Tool in a Se-mantic Knowledge Base.” In Human Language Technology Challenges for Computer Science and Linguistics: 5th Language and Technology Conference, LTC 2011 Poznań, Poland, Novem-ber 25-27, 2011 Revised Selected Papers, ed. by Zygmunt Vet-ulani and Joseph Mariani. Lecture Notes in Computer Science 8387. Cham: Springer, 356-66. doi: 10.1007/978-3-319-08958-4_29

Mazzocchi, Fulvio. 2018. “Knowledge Organization Sys-tem (KOS): An Introductory Critical Account.” Know- ledge Organization 45: 54-78.



McCourry, Maurine. 2015. “Domain Analytic and Domain Analytic-Like, Studies of Catalog Needs: Addressing the Ethical Dilemma of Catalog Codes Developed with Inadequate Knowledge of User Needs.” Knowledge Or-ganization. 42: 339-45.

Neelameghan, A. and G. J. Narayana. 2013. Concept and Ex-pression of Time: Cultural Variations and Impact on Knowledge Organization. New Delhi: Ess Ess.

Nietyksza, Maria. 1971. Ludność Warszawy na przełomie XIX i XX wieku. Warszawa: Państwowe Wydawnictwo Nau-kowe.

NUKAT. 2018. “Jhp KABA - Zespół Konsultacyjny, redaktorzy dziedzinowi”. https://centrum.nukat.edu. pl/pl/poznaj-nukat/zk-jhp-kaba

Oikarinen, Teijia and Terttu Kortelainen. 2013. “Chal-lenges of Diversity, Consistencyand Globality in Index-ing of Local Archeological Artifacts.” Knowledge Organi-zation 40: 123-35.

Ørom, Anders. 2003. “Knowledge Organization in the Domain of Art Studies — History, Transition and Con-ceptual Changes.” Knowledge Organization 30: 128-43.

Pawlak, Małgorzata and Karol Sanojca. 2018. “Ewolucja architektury treści polskiej i niemieckiej serii bibliografii historii Śląska.” In Bibliografie specjalne: rozwój i otwartość, ed. by Bartłomiej Włodarczyk and Jadwiga Woźniak-Kasperek. Warszawa: Oficyna Wydawnicza ASPRA-JR, 77-81.

Perks, Robert and Alistair Thomson, ed. 2016. The Oral History Reader. London: Routledge.

Petras, Vivien, Ray R. Larsonand Michael Buckland. 2006. “Time Period Directories: A Metadata Infrastructure for Placing Events in Temporal and Geographic Context.” In Proceedings of the 6th ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL'06). New York: ACM, 151-60. doi: 10.1145/1141753.1141782

Pleskot, Patryk. 2012. “Mimetyzm intelektualny. Zbieżności metodologiczne między historiografią polską a fran-cuskim kręgiem «Annales» (1956–1989).” In Klio Polska: studia i materiały z dziejów historiografii polskiej XIX-XX wieku 6: 37-54.

Purgarić-Kužić, Branka. 2006. „Povijesni sadržaji i Univer-zalna Decimalna Klasifikacija u nacionalnoj i sveuäœi- lišnoj knjižnici.” Vjesnik Bibliotekara Hrvatske 49, no. 3/4, 118-31.

Rabinowitz, Adam, Ryan Shaw, Sarah Buchanan, Patrick Golden and Eric Kansa. 2016. “Making Sense of the

Ways We Make Sense of the Past: The PeriodO Project.” Bulletin of the Institute of Classical Studies 59, 42-55. doi:10. 1111/j.2041-5370.2016.12037.x

Schneider, Alex and Daniel Woolf, ed. 2011. Historical Writ-ing Since 1945. Vol. 5 of The Oxford History of Historical Writing. Oxford: Oxford University Press.

Sima, Qian. 1993. Records of the Grand Historian: Qin Dynasty, trans. Burton Watson. Records of Civilization, Sources and Studies 65. Hong Kong; New York: Renditions - Columbia University Press.

Smiraglia, Richard P. 2012. “Epistemology of Domain Analysis.” In Cultural Frames of Knowledge, ed. by Richard P. Smiraglia and Hur-Li Lee. Würzburg: Ergon, 111-24.

Smiraglia, Richard. 2015. Domain Analysis for Knowledge Organ-ization: Tools for Ontology Extraction. Waltham, MA: Chan-dos.

Tennis, Joseph T. 2003. “Two Axes of Domains for Domain Analysis.” Knowledge Organization 30: 191-95.

Thucydides. 2017. History of the Peloponnesian War, trans. Richard Crawley. Mineola, NY: Dover.

Vann, Richard T. 2018. “Historiography.” In Encyclopaedia Britannica. https://www.britannica.com/topic/historio graphy

Wang, Shengang. 2019. “The Intellectual Landscape of the Domain of Culture and Ethics in Knowledge Organi-zation: An Analysis of Influential Authors and Works.” Cataloging & Classification Quarterly 57, no. 4, 227-43. doi: 10.1080/01639374.2019.1614710

Włodarczyk, Bartłomiej. 2014. “Deskryptory formalne w Bibliotece Kongresu a tematy formalne w języku haseł przedmiotowych Biblioteki Narodowej i w języku KABA.” Przegląd Biblioteczny 82: 518-36.

Wrzosek, Wojciech. 1997. History, Culture, Metaphor: The Facets of Non-Classical Historiography. Poznań: Wydawnic-two Naukowe UAM.

Zou, Qing and Eun G. Park. 2011. “Modelling Ancient Chinese Time Ontology.” Journal of Information Science 37, 332-41. doi: 10.1177/0165551511406063

Żarnowska, Anna. 1985. Robotnicy Warszawy na przełomie XIX i XX wieku. Warszawa: Państwowy Instytut Wydawniczy.

Zuo Tradition: Zuozhuan: Commentary on the Spring and Autumn Annals. 2016. Stephen Durrant, Wai-yee Li, David Schaberg, trans. 3 vols. Seattle: University of Washington Press.


D. Lee. Hornbostel-Sachs Classification of Musical Instruments

72

Hornbostel-Sachs Classification of Musical Instruments† Deborah Lee

City, University of London, Department of Library and Information Science, Northampton Square, London EC1V 0HB, England, <[email protected]>

Deborah Lee is a visiting lecturer at City, University of London, where she leads the information organization module. She was awarded her PhD in library and information science in 2017 from City, University of London. Her thesis, entitled “Modelling Music: A Theoretical Approach to the Classification of Notated Western Art Music,” was supervised by Professor David Bawden. Her research interests include music classification, the the-ory and aesthetics of classification schemes, music as information and the pedagogy of cataloguing education. Deborah is also the Joint Acting Head of the Book Library and Senior Cataloguer at the Courtauld Institute of Art.

Lee, Deborah. 2020. “Hornbostel-Sachs Classification of Musical Instruments.” Knowledge Organization 47(1): 72-91. 73 references. DOI:10.5771/0943-7444-2020-1-72. Abstract: This paper discusses the Hornbostel-Sachs Classification of Musical Instruments. This classification system

was originally designed for musical instruments and books about instruments, and was first published in German in 1914. Hornbostel-Sachs has dominated organological discourse and practice since its creation, and this article analyses the scheme’s context, background, versions and impact. The position of Hornbostel-Sachs in the history and development of instrument classification is explored. This is followed by a detailed analysis of the mechanics of the scheme, including its decimal notation, the influential broad categories of the scheme, its warrant and its typographical layout. The version history of the scheme is outlined and the relationships between versions is visualised, including its translations, the introduction of the electrophones category and the Musical Instruments Museums Online (MIMO) version designed for a digital environment. The reception of Hornbostel-Sachs is analysed, and its usage, criticism and impact are all considered. As well as dom-inating organological research and practice for over a century, it is shown that Hornbostel-Sachs also had a significant influence on the bibliographic classification of music.

Received: 27 May 2019; Revised: 4 June 2019; Accepted: X June 2019

Keywords: Hornbostel-Sachs, scheme, musical instruments, classification

† Derived from the article of similar title in the ISKO Encyclopedia of Knowledge Organization Version 1.1 (= 1.0 plus details on electrophones and Wikipedia); version 1.0 published 2019-01-17, this version 2019-05-29. Article category: KOS, specific (domain spe-cific). The author would like to thank the anonymous reviewers for their useful comments, as well as the editor, Professor Birger Hjør-land, for all of his insightful comments and ideas.

1.0 Introduction The “Classification of musical instruments” by Erich M. von Hornbostel and Curt Sachs (from here onwards, short-ened to Hornbostel-Sachs) is arguably the most important classification system for musical instruments. In fact, the scheme has played a significant role in the development of the domain in which it was created: “organology,” which is the study of musical instruments. First published in German in 1914, Hornbostel-Sachs has seen modifications and changes throughout its history, but the scheme remains part of contemporary musical instrument discourse and practice. While designed primarily to classify physical collections of musical instruments, Hornbostel-Sachs was redesigned in the 2000s as an ordering system for digital information about instruments. Hornbostel-Sachs’ originality centres upon its method of categorising instruments, its desire to be universal in terms of musical culture, its borrowing of decimal notation from bibliographic classification and more.

Its reach spreads throughout organology, and arguably Hornbostel-Sachs is sometimes seen as being synonymous with instrument classification.

The article starts with an overview of the origins and history of Hornbostel-Sachs, placing the scheme within the context of general developments of musical instru-ment classification and discussing the original purposes of the scheme. Next, the mechanics of the scheme are ex-plored, including its basic structure, atypical notation, war-rant and typographical layouts, among other areas. An ex-ploration of the versions and translations of Hornbostel-Sachs follows, highlighting the significance of the English translation of the scheme to its dissemination and discuss-ing new versions of the scheme. This is followed by an account of the reception of Hornbostel-Sachs, which con-siders its usage and criticism and culminates in an exami-nation of how Hornbostel-Sachs’ reach stretches beyond organology and instrument collections and becomes a cor-nerstone of bibliographic classifications of music.



73

2.0 Origins and history 2.1 Instrument classification up to the late

nineteenth century It is important to consider how Hornbostel-Sachs fits into the history of musical instrument classification. In the pre-Medieval eras, key instrument classification ideas came from the Old Testament of the Bible (especially Psalm 150), An-cient Greek ideas (in particular, works by Aristotle and Bo-etheius), and the Roman-era treatise of Cassidorus (Kart-omi 1990). In the Medieval and Renaissance periods, discus-sion centred on particular treatises, including those by Grocheo, Virdung, Zarlino, Praetorius and Mersenne (Kart-omi 1990). Some theories dominated multiple time periods; for instance, Kartomi (1990) claims that all writers on musi-cal instruments in the sixteenth and seventeenth centuries still referenced the Greek or Roman models of instrument classification. Ramey (1974) suggests that rather than a con-tinuously evolving discourse about instruments, the devel-opment of instrument classifications and theorisation of in-struments remained static from the seventeenth century for two hundred years. DeVale (1990) goes further still: she sug-gests that aside from adding the brass category, the basics of instrument classification in the western world were fun-damentally the same from Cassidorus’ scheme in the sixth century through the next 1,300 years. 2.2 Mahillon and the road to Hornbostel-Sachs It is at this juncture that a seismic change took place. Ma-hillon’s scheme (and corresponding catalogue) for the Conservatoire royal de Musique de Bruxelles was pub-lished in 1880 (Mahillon 1880; Jairazbhoy 1990a), and rad-ically altered the fabric of instrument classification in the western world. Mahillon’s scheme took the revelatory ap-proach of dividing the population of musical instruments into four, not three, categories, as had been the case for hundreds of years. Furthermore, the top-level categorisa-tion in Mahillon’s scheme divided instruments by how the sound was activated, rather than how the instrument was played. For a description of the categorisation of instru-ments including the categories used in Hornbostel-Sachs, see Section 3.1. Hornbostel- Sachs uses and expands Ma-hillon’s classification from thirty years earlier (Kartomi 1990), therefore, perpetuating the radical changes of Ma-hillon’s scheme. So, Mahillon’s scheme is the direct parent of Hornbostel-Sachs, and both these schemes are reactions to the prevalent trends in instrument categorisation that had developed up until the late nineteenth century.1

2.3 The germination of Hornbostel-Sachs Hornbostel-Sachs was developed in the early twentieth century by Austrian and German music theorists and scholars, Erich von Hornbostel and Curt Sachs (Katz 2001; Brown 2001). The scheme was first published in German in 1914 with the title of the scheme given as “Sys-tematik der Musikinstrumente” (Hornbostel and Sachs 1914). Note that this article uses the title of the scheme “Hornbostel-Sachs Classification of Musical Instru-ments,” which places the authors’ names in hyphenation in the title of the classification scheme, rather than just the translation “Classification of Musical Instruments.” As a further justification for using this format of the name, “Hornbostel-Sachs Classification of Musical Instruments” follows (with one exception) the formulation of the scheme’s title as found in an article about the scheme by Sachs in 1914, which calls it “Hornbostel-Sachs’sche Klas-sifikation der Musikinstrumente” (Sachs 1914, 1056). As well as the schedules, the 1914 scheme includes a detailed introduction, which explains the design of the scheme and outlines what it was trying to achieve, and this introduction is an important source in organology in its own right. 2.4 The purposes of Hornbostel-Sachs The authors of Hornbostel-Sachs had clear ideas about the users and purposes of their scheme. Hornbostel and Sachs (1961) were designing their scheme for musicolo-gists, ethnologists and curators of ethnological collections and cultural history. So, Hornbostel-Sachs was designed to be a scheme for theoretical and for practical purposes. We can also say that it is a knowledge organization system pri-marily designed for organising artefacts as opposed to mentefacts, using terminology used by the Classification Research Group (Gnoli 2018b). Note, organology is not mentioned in the introduction to Hornbostel-Sachs, as this term was not in common use in the early twenthieth cen-tury (see DeVale (1990) and Kartomi (1990) for the history and boundaries of organology as a domain). Another pur-pose given for the classification scheme (Hornbostel and Sachs 1961) is that it encourages researchers to find new links between instruments; so, Hornbostel-Sachs is ful-filling one criterion of being a scientific classification by enabling new knowledge to be created through classifica-tion (using the term “scientific classification” as way of describing a knowledge organization system created from within a domain (Mai 2011; Hjørland 2008; Lee, Robinson and Bawden 2018).

It is important to ask what problems Hornbostel-Sachs was attempting to solve. A key issue involves the culture of the knowledge being classified, and Hornbostel and Sachs (1961, 5) suggest that a classification that suits “one



74

era or nation may be unsuitable as a foundation for the instrumental armoury of all nations and all times.” Fur-thermore, one of the issues that the authors (Hornbostel and Sachs 1961) had with Mahillon’s scheme was their be-lief that it was led by European instruments. In contrast, Hornbostel and Sachs were attempting to create a classifi-cation that removed instruments from their corresponding cultures (Koch and Kopal 2014). Hornbostel and Sachs’ attempts to create what DeVale (1990, 8) delineated as a “cross-cultural system” (in opposition to a “culture-spe-cific system”), and this was part of a general move in or-ganology from the late nineteenth century onwards.

However, this brings to the fore questions about univer-sality and musical instrument classification. The results of Hornbostel and Sachs’ efforts to be cross-cultural in cover-age can be seen, for example, in decisions not to privilege instruments such as the violin or piano, which are especially associated with western art music. However, it is acknowl-edged that though this was their aim, Hornbostel and Sachs had their own specific temporal and cultural background that cannot be ignored. The disjuncture between Hornbos-tel-Sachs’ context-neutral design and the cultural-historical context of its creation (“wissenschaftshistorischen Entste-hungszusammenhängen”) is articulated by Koch and Kopal (2014, 301). In DeVale’s (1990) continuums for analysing or-ganological classifications, Hornbostel-Sachs could be con-sidered as mostly exogenous, arguably like any cross-cultural classification system; the authors exist outside of the cul-tures of the majority of instruments covered by the scheme, as an inevitable result of the scheme covering a variety of different musical cultures. Therefore, for many classes in their scheme, Hornbostel and Sachs are imposing classifica-tion on the instruments and the cultures that those instru-ments represent. Furthermore, recent knowledge organiza-tion discourse acknowledges the conceptual issues with uni-versality as a desired attribute of a knowledge organization system, and the blurred definitions of the concept (see, for example Szostak 2014), as well as increasing awareness that neutrality is not an attainable (or even always a desired) goal. So, there is a tension between Hornbostel and Sachs’ ambi-tion of writing a cross-cultural scheme, and the reality of cross-cultural instrument classification delivered through a single scheme.

Another issue that Hornbostel-Sachs tries to resolve concerns the historical three-category system of classify-ing instruments. The authors describe the three traditional categories of instruments as “illogical” and “inadequate,” and are complementary about the four categories used in Mahillon’s scheme (Hornbostel and Sachs 1961, 6) (A full discussion of the four categories is found in Section 3.1. of this article). So, the obvious question is why Hornbostel and Sachs did not just extend or develop Mahillon’s scheme? One reason is that while Hornbostel and Sachs

utilised Mahillon’s four categories, they found issues with the logic used within each of Mahillon’s categories. There-fore, another purpose of Hornbostel-Sachs was to provide what the authors considered to be a logical division and structure of musical instruments, within a four-category system. 3.0 The mechanics of the scheme This section explores the mechanisms of Hornbostel-Sachs as a classification scheme. The 1914 version will be used as the baseline scheme, in its 1961 English translation, unless otherwise stated. 3.1 Four categories A revolutionary aspect of Hornbostel-Sachs occurs at its highest level: the division of the universe of musical instru-ments into four categories. Until the late nineteenth century, western classifications of instruments were organised into three broad categories: wind, strings and percussion. Ma-hillon’s 1880 scheme instead had four, not three, categories based around how the sound was made, and these categories were named “instruments autophones,” “instruments à membranes,” “instruments à vent” and “instruments à cordes” (Mahillon 1880).

This quadrivium became the basis of Hornbostel-Sachs, albeit with some changes in nomenclature. First, Hornbostel and Sachs takes the neologism found in Mahillon’s scheme of naming a category of instrument using “method of sound production” plus “phones,” and applies it to the other three categories; for example, “instruments à cordes” in Ma-hillon’s scheme becomes “Chordophone” in Hornbostel-Sachs. Note that the German terms for the category names are given, so “Chordophone” is a plural in the 1914 original German, which becomes “chordophones” once translated to English. Second, Hornbostel and Sachs prefer the term “idiophone” rather than Mahillon’s “autophones”; the au-thors were concerned that a category entitled “autophones” might be confused with automatic instruments (Sachs 1914). This decision was based on research by Sachs pub-lished in 1913 (Sachs 1914; Hornbostel and Sachs 1961). Figure 1 shows the progression of the main categories over time.

Note that there is some debate about the novelty of these four categories, which in turn could affect the per-ception of Hornbostel-Sachs as a disruption of classifica-tory norms: scholars have commentated that Mahillon’s and Hornbostel-Sachs’ four-category system appears to match the classification espoused in a fifth-century Indian treatise, “Nātyasāstra,” attributed to Bharata (Jairazbhoy 1990a; Heyde 1977). Furthermore, Jairazbhoy (1990a) pos-its that Mahillon would have been aware of “Nātyasāstra”



75

and describes anomalies that strongly infer that Mahillon consciously borrowed the Indian four-category system. So, although the radical four-category system of Hornbostel-Sachs is credited to Mahillon, the origins of a four-cate-gory system is not (only) a nineteenth century, European invention.

The four-category system did, however, break the con-tinuum of European instrument classification in a number of ways. First, what could be called the “characteristic of division” is different in the four-category systems of Ma-hillon and Hornbostel-Sachs from traditional three-category schemes. For example, both a gong and a drum are struck, so in a traditional three-part system are considered “percus-sion;” however, in the four-category system, the fact that the gong’s sound is produced by the solid material of the gong itself vibrating and the drum’s sound is produced by the stretched membrane of the drum vibrating means they would be placed in different classes (classed in membrano-phones and idiophones respectively in Hornbostel-Sachs). Second, the four-category system gives much more space to what in older categorizations would be called “percussion” instruments, and notably in Hornbostel-Sachs, sees the per-cussion-equivalents appear earlier in the order of categories. The prominence of idiophones and membranophones in Hornbostel-Sachs is no accident. In systems designed pri-

marily for western art music, the percussion instruments have far less importance than in the whole universe of music cultures; categorisng based on sound-production means that the privileging of instruments from one culture at the ex-pense of other cultures is reduced, thus enabling Hornbos-tel and Sachs’ universal intentions. However, although Hornbostel-Sachs is noted for its four categories, a fifth cat-egory for electrophones becomes the standard structure in later years—see Section 4.3. 3.2 Notation Hornbostel-Sachs is remarkable as it uses a decimal nota-tion (Gnoli 2018b, Section 3.1), and this feature alone makes it significantly different from its Mahillon parent-age. While attributed by the authors of Hornbostel-Sachs as being a Dewey Decimal Classification (DDC) notation (Hornbostel and Sachs 1961), a close look at Hornbostel-Sachs makes it clear that it is not exactly DDC that has been used. Gnoli (2006) states that Hornbostel-Sachs uses the European version of DDC, which is an authorised ver-sion of DDC and the precursor to the “Universal Decimal Classification” (UDC) (Gnoli 2006). There are a number of ways in which Hornbostel-Sachs adopts UDC notation rather than pure DDC. For example, the four base catego-

Figure 1. The top-level categories in Hornbostel-Sachs and its antecedents.



76

ries are given as “1,” “2,” and so on; yet, if this used DDC’s decimal notation, extra zeros would be added so that these categories would have the notation “100,” “200,” and so on. In addition, the presence of a period every three digits also marks out Hornbostel-Sachs as adopting Euro-pean/UDC notation; for instance, Hornbostel-Sachs has “211.212.2” for “sets of cylindrical drums” (Hornbostel and Sachs 1961), which would have been written as “211.2122” (or similar with apostrophes) in DDC’s nota-tion.

Hornbostel-Sachs’ notation works by each new level of classification adding an extra digit to the right-hand side of the notation (see Gnoli 2018b, Section 3.1, for more infor-mation on how decimal notations function). For example, “4” represents aerophones, “41” free aerophones, and “411” displacement free aerophones (which includes the whip or sword-blade). There is also something interesting about the digits selected by Hornbostel and Sachs. For ex-ample, the aerophones are divided into three main types: free aerophones (41), wind instruments proper (42), and trumpets (43). Note that the digits “1,” “2” and “3” have been selected; this is in contrast to what we might expect to find in other decimal systems where three digits selected across the range from “1” to “9” might be used instead.2 For an example of Hornbostel-Sachs notation, see Figure 2, which presents a selection of classes from aerophones with their corresponding notations. The same design can be seen at every level of Hornbostel-Sachs; frequently, only the digits “1” and “2” are used, and there are very few occurrences of digits over “4.” The impact of this choice is that it diminishes Hornbostel-Sachs’ hospitality: there is no room to insert a new category between existing catego-ries in future versions, at least in a way that keeps the new category at the same level of notation as its siblings. So, when Montagu (2009) wishes to add a new high-level group

Figure 2. Selection of classes from aerophones demonstrating decimal classification (Hornbostel and Sachs 1961). Note, only the first three levels are shown.

for the half-spike lute, which conceptually fits between the spike-lutes (321.31) and necked lutes (321.32), he is faced with a notational problem. Montagu’s (2009) solution was to add a new class of 321.31.5, to sit between 321.31 and 321.32. This destroys the symmetry of having only 3 num-bers between periods; also, the choice of “5” appears to be representing thirty-one-and-a-half, rather than a class equal in hierarchy to 321.31 and 321.32. Therefore, adding new classes is unideal in Hornbostel-Sachs due to its particular application of decimal notation.

Another and related feature of Hornbostel-Sachs’ no-tation is its expressivity. For instance, the class with nota-tion “1” (idiophones) has only one digit and is a very broad category, whereas “111.141” (castanets) has six digits and represents a specific type of instrument (though as will be discussed below, not an actual instrument). However, as the classification scheme does not have the same number of hierarchical levels between category and type of instru-ment for each type of instrument, as discussed in more detail in Section 3.4, this means that the notation cannot be fully expressive; for example “421.121.12” is a category with eight digits representing a specific type of end-blown flute, while the three-digit “413” for “plosive aerophones” also represents a specific type of instrument. So, while it is generally true that the number of digits represents some broad idea of where you are within the hierarchy, there is some variation for different areas of the scheme. 3.3 Arrangement within categories Hornbostel-Sachs (1961) has a different order of knowledge within each of its four main categories. The authors selected the most appropriate divisions for each category rather than consistently apply the same criteria or order of these criteria across each of the four main categories (Hornbostel and Sachs 1961). Furthermore, Hornbostel and Sachs are concerned with placing too much emphasis on method of playing as a main dividing principle, which is the basis of Mahillon’s scheme: for example, if playing method is the primary way of dividing chordophones, then the plucked violin and bowed violin would go in very dif-ferent places, yet they are the same instrument (Hornbostel and Sachs 1961). Gnoli (2006) summarises the different orders within the four main categories as follows: while chordophones and aerophones are mostly concerned with morphology, the playing technique largely governs the or-dering of the idiophones and membranophones categories. The structure of the first two levels within each of the four main categories is shown in Figure 3, Figure 4, Figure 5 and Figure 6, and these figures highlight the inconsistencies in structure between the classes. Kartomi (1990) suggests that Hornbostel and Sachs’ reason for forgoing logical division was a pragmatic choice, where the complexities of reality



77

Figure 3. The first two levels within the idiophones class, extracted from Hornbostel and Sachs (1961).

Figure 4. The first two levels within the membrano-phones class, extracted from Hornbostel and Sachs (1961).

Figure 5. The first two levels within the chordophones class, extracted from Hornbostel and Sachs (1961).

Figure 6. The first two levels within the aerophones class, ex-tracted from Hornbostel and Sachs (1961).



78

win out over classificatory niceness. However, the incon-sistency in ordering within categories is given as one of the main criticisms of Hornbostel-Sachs (for example, see Wachsmann et al. 2001).

These points give some insight into the philosophical foundations of Hornbostel-Sachs. This suggests that Horn-bostel-Sachs adopts an ontological approach, where the phenomena (the instruments) are considered to be the cen-tre of the classification scheme, and decisions about the hi-erarchical levels used in the scheme are driven by what is found in the real world of instruments.

The authors of Hornbostel-Sachs are particularly elo-quent about their choices for the terminology of knowledge-levels within each category. Hornbostel and Sachs (1961) decided not to formally label the levels within each category, despite their comment that labels of strata are used in biological classification and in Mahillon’s scheme; however, Hornbostel and Sachs (1961) do suggest infor-mal names for these levels, suggesting that the idiophones and so on would be called classes, followed by sub-classes, orders and sub-orders. The authors’ comments mention-ing biological classification are interesting as they could be read as a link between organology and other scientific clas-sifications. Furthermore, the deliberate omission of offi-cial terms for the levels within the hierarchies could be viewed as an expression of the confusion within music classification about how to appellate the chains between broad instrument categories and individual instruments. For example, issues about which levels in the chain are cov-ered by the amorphous term of “instrument family” are explored conceptually by Lee (2017c) in her discussions about string ensembles, and in practical terms in a DDC working paper (Dewey Decimal Classification 2016).

3.4 Individual classes The size of the chain in Hornbostel-Sachs between broad category (for example, “idiophones,” “membranophones”) and lowest level class (for example, “slide trumpets,” “dou-ble-skin stationary drums with friction-cord”) varies across the scheme. For example, “friction drum with whirling stick” is at 232.2, showing only four levels of hierarchy, and “free kazoos” is at 241, showing only three levels of hier-archy; conversely, the “without tuning noose” mono-heter-ochord music bow with resonator is at 311.121.221, show-ing nine levels of hierarchy. An example of the hierarchy leading from chordophones to the class 311.121.221 can be seen in Figure 7. Figure 7 also demonstrates how the deci-mal notation adopted by Hornbostel-Sachs makes it simple to see the hierarchical pedigree of any class; for example, one glance at the number 311.121.221 shows that it con-tains 311.121 (mono-hetereochord musical bow), meaning that 311.121.221 must be a mono-heterochord musical bow, because it has the (great grand-) parent class of 311.121 included within it.

These lowest levels of classes are not titled by names of specific instruments. Instead, the lowest-level classes have titles that are the shared characteristics of instru-ments, which would reside in those classes. Specific instru-ments are given as selective examples, such as the hade, African lyre, violin, European flute, Ocarina, and so on. For example, class 321.322 is entitled “necked box lutes or necked guitars,” with a note stating “violin, viol, guitar.” This list only contains selected examples, and any instru-ment considered a necked box lute of necked guitar would be classed here, such as violas, cellos or the double bass (to give some examples important to western art music).

Figure 7. Example of hierarchy in 311.121.221 (Hornbostel and Sachs 1961). Note, classes have been omitted which are not direct descendants of 311.121.221.



79

These examples and notes will be examined in more detail in Section 3.7, which discusses typographical layout. 3.5 Coverage and warrant The “necked box lutes or necked guitars” example illustrates a number of important points about the coverage of Horn-bostel-Sachs. First, the authors’ intention of being culturally universal is illuminated by this example, as even the typically western instruments in this class are still only given as exam-ples rather than the title of the class. Second, this example shows how Hornbostel-Sachs does not distinguish between current and obsolete instruments; viols are usually associ-ated with music of the seventeenth century and earlier and were largely superseded by the violin, viola and cello, while violins and guitars are popular (in specific cultures) in the twenty-first century. This fits with Hornbostel-Sachs’ phi-losophy to be for all times, as laid out in the scheme’s intro-duction (Hornbostel and Sachs 1961) and also seen in the tensions presented by each class representing both the cur-rent instrument and its evolutionary progression to get to that form (Gnoli 2006). Furthermore, this treatment of temporal existence of instruments also fits into Hornbostel-Sachs’ purpose of organising collections of instruments, as instrument collections usually contain many historical, “su-perseded” instruments.

Literary warrant is another important aspect of under-standing Hornbostel-Sachs, and literary warrant is taken as a broad term to include all types of documents including objects such as musical instruments (Barité 2018). Horn-bostel and Sachs discuss whether instruments need to exist to be included: “we have refrained from providing a sub-division containing no existing representative” (Hornbos-tel and Sachs 1961, 10). In other words, there is a literary warrant for any class to be included in Hornbostel-Sachs. This contrasts with Mahillon’s scheme (on which Horn-bostel-Sachs is based), which includes categories for in-struments which had not yet been invented (Jairazbhoy 1990b, 82-83).

However, the question of literary warrant is not quite this straightforward. The Hornbostel-Sachs schedules sug-gest that the scheme itself is less clear-cut than is implied in its introduction. First, there are a very small number of classes where the example or note has the word “un-known,” suggesting that either specific instruments are un-known or the geographic location where such instruments are found is unknown. Examples include “132.1 (Individ-ual) friction sticks” (under Friction sticks) and “131.1 (In-dividual) friction sticks” (under friction plaques).

Second, there is one class, “421.121.311 with fixed stopped lower end” (under stopped side-blown flutes) which has the note “Apparently non-existent.” It is not ex-plained whether there are just no extant exemplars of that

instrument, or there is no evidence that the instrument ever existed. At least some of these examples, especially the friction sticks, might be explained by the authors’ (Hornbostel and Sachs 1961) caveat in the introduction to the scheme: sometimes they assigned classes to simpler versions of a known instrument, because they assume this earlier and simpler version existed, even if they do not have evidence. So, the literary warrant of Hornbostel-Sachs is not just instruments that were known to have def-initely existed, from the temporal perspective of 1914; in-stead, the literary warrant of Hornbostel-Sachs also in-cludes instruments that were thought to have existed, from the temporal perspective of 1914. 3.6 Extensions and alternatives Hornbostel-Sachs includes a number of extensions. These extensions could be considered number-building, or even loosely as an analytico-synthetic feature of the scheme. Furthermore, some alternatives are also offered.

There are two broad types of extension in Hornbostel-Sachs. The first type involves optional additions to existing classes, which are different depending on the broad cate-gory of instruments. These additions are found at the end of each category, and are suffixes to the main classes. For example, you can add “with membrane glued to drum” to any class within membranophones (Hornbostel and Sachs 1961). However, you can only add it to a class in the mem-branophones category; clearly, this addition would make little sense to a class in chordophones or aerophones. To add this suffix, a dash is added to the regular class, and it is possible in some cases to add multiple additions. The purpose of these extensions is to provide more detail to existing classes. To some degree these extensions could be considered a light sort of synthesis, at least within the uni-verse of any individual broad category of instruments such as membranophones.

The second type of extension involves building a new class from two or more existing classes. The introduction to Hornbostel-Sachs (1961) gives an example of the mod-ern, western orchestral trombone, which has slides and valves; in Hornbostel-Sachs, the slide trombone is found at 423.22 and the valve trombone is found at 423.23. Hornbostel-Sachs (1961) says that this instrument could be represented using both classes, with a plus between the two notations (423.22 + 423.23). A notational short-cut is also offered: 4232.2 + 3. This shorthand notation works by using the period to indicate the division between the digits that are being repeated (in this case, 4232) and those digits which are not; so, this class reads 423.22 + 423.23, with the user alerted to the repeat of “4232” by the posi-tion of the period. Changing the position of the period is an interesting variation on decimal notation, and is not



80

seen in schemes such as DDC. This type of extension sees composite instruments intellectually represented as a com-bination of two types of instruments. This could be con-sidered as faceting, where the composite notation repre-senting the composite instrument could be considered as a complex class created from the simple classes of the in-itial instruments. This is taken further by the examples of bagpipes given in Hornbostel and Sachs (1961), and dis-cussed in Ghirardini and Gnoli (2005): using the decimal point, brackets and colons (for example, 422-62:22 for a reed instrument with flexible air reservoir with exclusively clarinet pipes, or 422-62 : .2]1 for set of reedpipes with flexible air reservoir with bagpipe of oboe), this quasi fac-eting can be used to specify the way sound is produced in different parts of the instrument using the existing sched-ules and extensions to build a more specific notation than is listed in the schedules.

There is a third, and vaguer type of extension. For ex-ample, the introduction (Hornbostel and Sachs, 1961) says that adding new subdivisions, principles or classes is also acceptable. In other words, the classifier is free to change and extend the scheme at will. Consequently, Hornbostel-Sachs is theoretically infinite in its coverage, although the limitations of decimal notation in base ten mean that the scheme cannot be infinitely hospitable in terms of how this infinite universe is notated.

Hornbostel-Sachs also permits alternatives. For in-stance, the bagpipes example (Hornbostel and Sachs 1961) also shows how suffixes in the aerophones class are made part of the main class, thus altering the order of elements. This flexibility would be useful for curators and others se-lecting the most appropriate arrangement for their specific collection. However, this does have an impact on univer-sality, as it suggests Hornbostel and Sachs (1961) did not intend for their scheme to be used in the same structure at different institutions and that the authors were not posit-ing a singular order of instruments. 3.7 Typographical layout It is important to consider how the information within a scheme is structured and labelled, as it illustrates the au-thors’ structural intentions and is often used to communi-cate these intentions to the schemes’ classifiers. The Horn-bostel-Sachs typography is particularly interesting as dif-ferent layouts of information are adopted in the German original, English translation of 1961, English reprint of 1992 and 2011 updated version. (For more information about these versions, see Section 4.) This asks important questions about typographical meaning and its transmis-sion across versions of schemes.

Hornbostel-Sachs contains four types of information in the main schedules: the notation, title of class, descrip-

tion and notes. The notes are varied, often containing ge-ographic information about where the instrument is found (for example, “China und Vorderindien” (Hornbostel and Sachs 1914, 563)) or examples of specific instruments (for example, “Violine, Gambe, Guitarre” (Hornbostel and Sachs 1914, 580)). In the original German version of the scheme (Hornbostel and Sachs 1914), the scheme is laid out as a table, with the three columns labelled as follows: Klassifikation (which contains the notation and title of class), Charakteristik and Beispiele. The table is presented in landscape form.

The tabular layout is not followed through to subse-quent versions (such as 1961, 1992 and 2011), nor are there any labels for the different types of information; instead, the differentiation between notation/class and other types of information is presented using typefaces and punctua-tion. The 1961 and 2011 versions use bold typeface for the notation and class titles, with roman typeface for the “Charakteristik,” and italic typeface for the equivalent of the “Beispiele” (The 1961 and 2011 versions are typo-graphically very similar, apart from the significant spacing between “Charakteristik” and “Beispiele” in the 1961 ver-sion, and completely different family of typefaces used in the 1961 and 2011 versions). The formatting in the 1961 and 2011 versions makes it relatively easy to distinguish the different types of information, even without Hornbostel and Sachs’ (1914) labels or tabular layout.

The 1992 reproduction adopts a different typographical layout from the original 1961 English translation. In the 1992 version, the scheme is presented in two unmarked columns of text, with the notation in the left column and the other types of information in the right column. Class titles are in bold, upper-case letters or italics, depending on their hierarchy. The equivalent of the “Charakteristik” is in roman typeface, usually preceded by a colon. However, the equivalent of the “Beispiele” is also given in roman type-face. Sometimes the Beispiele-equivalent is preceded by an m-dash, with an introductory phrase “found in” for geo-graphical examples, and with other examples the Beispiele-equivalent is displayed in parenthesis. Furthermore, some of the examples are located in a different place within the description of a class, as compared to the German original. The overall effect of the 1992 typographic layout arguably makes it more difficult to delineate the different types of information than other versions. This could be interpreted as a sign that the strong divisions found in the 1914 version between the different types of information that constitute the scheme, are not considered a core tenet of Hornbos-tel-Sachs (or at least, were not considered core by those responsible for the 1992 version). The different types of information that make up the presentation of a classifica-tion scheme could be considered key parts of the “verbal plane” of a knowledge organization system (where the ver-



81

bal plane is an intrinsic part of a classification scheme and separate but related to the structural elements found in the notational plane (Gnoli et al. 2011)). These verbal planes are an important part of the knowledge organisation of databases (Gnoli et al. 2011), which is especially interesting when considering the intended use of the 2011 version of Hornbostel-Sachs (see Section 4.4).

Another important factor is to examine the typographical representation of Hornbostel-Sachs’ hierarchy. The original German scheme used indentation to indicate hierarchy; for example, 322.21 is set to the right of its parent class 322.2, but to the left of its child, 322.211. However, the difficulties of presenting indentation within a column of a table may be the reason that only the first few levels of the hierarchy are indented. The 1961 and 2011 versions use a subtle form of indentation to indicate hierarchy, using the differences in length of the decimal notation to indent the class names. The 1992 scheme does not use indentation to represent hi-erarchy, which asks a question about whether hierarchy was considered an important aspect of Hornobostel-Sachs to the editors of this version. All four versions also make some use of changes in typeface to represent different levels of hierarchy. The German original uses a bold weight for the four main categories, and italic typeface for the fourth level; the 1961 and 2011 versions only differentiate the first level from all the other levels, and do this using upper-case letters and type size; the 1992 English version uses bold weight for the four main categories, upper-case letters for the next level down, and then italic typeface for all other levels. In conclu-sion, hierarchy is represented typographically in Hornbos-tel-Sachs in various ways and comparing the different ver-sions illustrates different approaches to the importance of visually representing the hierarchy of Hornbostel-Sachs, such as the 1992 version forgoing the communication of hierarchy through indentation. Furthermore, these varia-tions between different versions indicate that the visual rep-resentation of hierarchy is not deeply imbued within Horn-bostel-Sachs, as it is not transmitted in a consistent form be-tween versions. 3.8 The introduction to Hornbostel-Sachs Hornbostel-Sachs includes a substantial introduction writ-ten by its authors. The introduction makes up a sizeable component of the scheme; for example, there are eleven-and-a-bit pages of introduction in the 1961 edition, which is the same size as the scheme itself. The introduction in-cludes the following: an outline of why a systematic classi-fication is needed and the purposes of the scheme; ideas about being a classification for all cultures and this as a driving force behind the creation of the scheme; the prob-lems of the incumbent three-category system; details about, and a critique of, Hornbostel-Sachs’s direct prede-

cessor, the Mahillon scheme; an account of the structure of Hornbostel-Sachs and explanations for some of its structural features; an explanation of Hornbostel-Sachs’ notation; number-building and alternatives. So, as well as being a practical guide to using the scheme, the introduc-tion also serves as Hornbostel-Sachs’ manifesto.

The introduction to Hornbostel-Sachs is such an im-portant source in its own right that not only is it quoted by numerous commentators on instrument classification, but the English translation of the introduction is also re-printed in various “Grove” (the prominent encyclopaedia of music) resources. The introduction to Hornbostel-Sachs appears in the 1980 New Grove Dictionary of Music and Musicians and remains in the current version of this re-source, as an appendix to the entry on the classification of instruments (Wachsmann et al. 1980, 2001), as well as ap-pearing in the New Grove Dictionary of Instruments (Wachs-mann et al. 1984). This indicates the centrality of Horn-bostel-Sachs to the development of organology; this is fur-ther discussed in Section 5.3, which places this discussion in the context of the impact of Hornbostel-Sachs. 4.0 Editions, updates and revisions 4.1 Introducing updates Hornbostel-Sachs was first published in 1914 under the ti-tle Systematik der Musikinstrumente: ein Versuch, within the German journal Zeitschrift für Ethnologie: Organ der Berliner Gesellschaft für Anthropologie, Ethnologie und Urgeschichte. In other words, this seminal classification scheme was first published in a journal for anthropology and ethnology, ra-ther than being in a musical journal. Furthermore, the term “ein Versuch” suggests it was intended as a discussion doc-ument rather than the absolute version of a classification scheme (Montagu 2009). The publication language and subject area are seen by some commentators to have had a negative effect on Hornbostel-Sachs. Baines and Wausch-smann ([Translator’s introduction], Hornbostel and Sachs 1961), suggest that the German language and the subject matter of the home journal hindered the accessibility of the scheme in its first fifty or so years. Similarly, Montagu (2009) suggests that discussion about the proposed classi-fication scheme was curtailed by the unfortunate timing of the scheme’s publication in 1914, when Europeans had other urgent matters to consider aside from the classifica-tion of instruments.

After initial publication, Hornbostel-Sachs was subse-quently adapted and altered. In fact, adapting Hornbostel-Sachs appears to have been prevalent even from the first years of the scheme: for example, Montandon produced an adaptation of the scheme in 1919, just five years after its original publication (Dournon 1992).



82

4.2 Translations Hornbostel-Sachs was translated into English for an issue of the “Galpin Society Journal” in 1961. The translators, Baines and Wachsmann, describe the motivation for their translation: the original German Hornbostel-Sachs is often quoted by those working with or researching instruments, and has not been superseded in terms of usage ([Translator’s introduction], Hornbostel and Sachs 1961). Interestingly, Baines and Wachsmann ([Translator’s introduction], Horn-bostel and Sachs 1961) acknowledge that modifications have been made to Hornbostel-Sachs between 1914 and 1961, but they choose to translate the original 1914 text rather than any modified version. This is all the more remarkable considering the development of a new class of electro-phones in the 1930s (see Section 4.3). The translators ([Translator’s introduction], Hornbostel and Sachs 1961) were concerned about students having access to original classification schemes as a source; this situates the original 1914 scheme as an important document in its own right, of such value that it should be studied in its frozen form nearly fifty years later. This shows the pre-eminence of Hornbos-tel-Sachs within the organology community, and also illumi-nates the versioning (to use the terminology of Tennis 2010) of the scheme by placing (retrospectively) the 1914 original as a distinct and self-contained document rather than part of a continuum of minor adaptations and small updates.

In 1992, the scheme appeared in its English translation as part of an anthology of “reference aids” within an eth-nomusicology textbook (Hornbostel and Sachs 1992). Again, this inclusion places Hornbostel-Sachs as a cultural work in its own right, presented as a canonical document of ethnomusicology. Whether the scheme appearing in a text-book of ethnomusicology as opposed to being available in a back issue of a key journal in organology would have had much of an impact on availability, will not be explored, but it is assumed that any availability issues for the 1961 edition was resolved once older issues of the Galpin Society Journal became available electronically. The 1992 reprint differs from the 1961 version in terms of layout, typography, pagi-nation and the absence of the translators’ introduction, but not in the contents of the scheme or authors’ introduction.

However, there are other translations of Hornbostel-Sachs: for instance, the translation into Italian of the scheme and its introduction by Guizzi (2002), as well as translations into languages such as Catalan, Finnish and Spanish mentioned by authors commentating on Horn-bostel and Sachs (Kartomi 2001; Montagu 2009). The translations are important to note when considering the universal intentions of Hornbostel-Sachs, and that trans-lations are one aspect of considering the impact of a clas-sification scheme (Lee 2015).

4.3 New categories: introducing electrophones A significant development sees the introduction of a fifth category of instruments. This category is first found in Gal-pin’s (1937) book about European musical instruments, un-der the title “electrophonic instruments.” Galpin’s (1937) book includes a discussion about instrument classification which discusses Hornbostel-Sachs and an outline of Gal-pin’s classification scheme. Furthermore, the chapters of the book are assigned to the broad categories of instruments including the new category “electrophonic instruments” (Galpin 1937). Galpin (1937, 30) acknowledges that this class is “entirely new and included here for the first time.” Galpin (1937, 30) defines electrophonic instruments as “in-struments in which the sound-waves are formed by oscilla-tions set up in electronic waves.” However, the scheme that Galpin (1937) presents and discusses is an updated version of Galpin’s own 1900 scheme, which appeared between Ma-hillon’s scheme in 1880 and Hornbostel-Sachs in 1914.

The first appearance of the fifth category as part of Hornbostel-Sachs appeared in 1940, in a history of musical instruments by Curt Sachs (1940). Sachs (1940, 455) states that there are “five main classes” of instruments, and there is a section for electrophones alongside the existing four cat-egories in the “Terminology” chapter of the book. The term “electrophones” is the typical title for instruments of this nature, and this is the term adopted by the MIMO ver-sion of Hornbostel-Sachs (see Section 4.4). However, the boundaries of electrophones can be ambiguous (Kartomi 1990). For example, Bakan et al. (1990) discuss distinguish-ing between “electrophones” and “electronophones” when talking about the classification of electronic music instru-ments. The presence of the fifth category for electrophones in the general Wikipedia article on Hornbostel-Sachs (Horn-bostel-Sachs 2019) could be seen as evidence of the ac-cepted norm of Hornbostel-Sachs being considered a five-category scheme.

Other new categories in Hornbostel-Sachs have also been suggested. For example, Olsen (1986) proposes a new fundamental category for sound produced by using the hu-man body as an instrument, called corpophones. Unlike electrophones, “corpophones” do not (yet) seem to have been adopted as a standard category. 4.4 MIMO: a new version of Hornbostel-Sachs? There is one resource that has strong arguments for being considered a truly distinct version of Hornbostel-Sachs, as opposed to just a new state (using the distinction found in Tennis (2010)). A new version of Hornbostel-Sachs was published electronically in the twenty-first century, for use by the Musical Instrument Museums Online (MIMO) pro-ject. The MIMO project’s purpose was to create “a single



83

access point to digital content and information on the col-lections of musical instruments held in a consortium of Eu-ropean museums” (Musical Instruments Museums Online 2011, 1); to fulfil this aim, a new version of Hornbostel-Sachs was created (Musical Instruments Museums Online 2011). According to the introduction to the scheme (Musical Instruments Museums Online 2011), the main purpose of this revision was to classify instruments that were invented since the 1914 scheme was published, such as electro-phones. From a knowledge organization perspective, this comment is insightful: the purpose of the MIMO revision was not to rethink the structure of instrument classification but predominantly to incorporate the new knowledge that had been generated since the scheme was originally created.

The MIMO version of Hornbostel-Sachs was created by the “MIMO Working Group for Classification and The-sauri,” chaired by Margeret Birley of the Horniman Mu-seum, London (Musical Instruments Museums Online 2011, 1). However, this version is closely related to another instantiation of the scheme: the revised version of Horn-bostel-Sachs by the organologist Jeremy Montagu (Musical Instruments Museums Online 2011). Therefore, to consider the MIMO version of Hornbostel-Sachs, it is imperative to

also consider its direct descendant. Montagu’s version was published in 2009, in the Polish music journal Muzyka.3 Montagu (2009) uses typographical features such as crossing out text and asterisks to indicate changes from the 1961 translation of the 1914 scheme, to his new version. Further-more, Montagu’s (2009) version builds on his previous re-search and modifications to Hornbostel-Sachs from earlier years, such as his work with Burton in 1971 (Montagu and Burton 1971). From this we can see that the MIMO version is a substantial revision, but its creation is part of a contin-uum of scholarship through the twentieth and twenty-first centuries, with Montagu a key creator of these develop-ments. See Figure 8 for a visualisation of the MIMO scheme and some of its antecedents, which shows the types of rela-tionships between the various instantiations.

Arguably the most significant change between the MIMO version and the original 1914 version of Hornbos-tel-Sachs is the addition and development of the electro-phones main category. The inclusion of electrophones was inevitable, and had been missing from the English transla-tion in 1961 and 1992. The significant stages between Gal-pin’s initial use of “electrophonic instruments” and the in-clusion of an electrophones as a fifth category in the

Figure 8. The relationships between MIMO version of Hornbostel-Sachs and its antecedents.



84

MIMO version are illustrated in Figure 9, although note this figure only contains highlights of this development. The schedules of electrophones in MIMO are based on the work of Maarten Quanten (Musical Instruments Mu-seums Online 2011).

The MIMO version also made structural changes in other categories. Examples given in the MIMO introduc-tion (Musical Instruments Museums Online 2011) include expansions and renaming of part of kettledrums, a differ-ent sort of division employed in the brasswinds and new categories in aerophones. Finally, the MIMO version was designed for a shared, digital environment and this has meant changes to the notation; for example, the abbrevia-tions in notation suggested by Hornbostel and Sachs in their introduction (Hornbostel and Sachs 1914) have been omitted. For more details about the developments en-sconced in the Montagu and MIMO versions, see the in-troductions to Montagu (2009) and MIMO (Musical In-struments Museums Online 2011).

The development of the MIMO version should not be seen as the end of the variations, amendments and version-ing of Hornbostel-Sachs. Adding new classes and structural

changes do not in themselves resolve issues of using a hier-archical structure, especially in an online age. For instance, Weisser and Quanten (2011), writing at the same sort of time as the MIMO revisions were published and dissemi-nated, argue for a different format and approach to Horn-bostel-Sachs. They (Weisser and Quanten 2011) do not con-sider forcing the classifier down a single path based on initial vibration is satisfactory for all instruments. So, like countless others before them, Weisser and Quanten (2011) suggest a new way of using Hornbostel-Sachs, with additions and amendments. Therefore, the MIMO version is not an end-point, as the amendments, modifications and rethinking of Hornbostel-Sachs keep on coming. 4.5 Governance of Hornbostel-Sachs As a postlude, it is interesting to briefly consider the own-ership and maintenance of Hornbostel-Sachs. The original scheme was published in a journal, and the important 1961 English translation was also published within a journal and then is republished as a book chapter in 1992. Later versions were published as papers by Montagu and then as

Figure 9. The relationships between MIMO version of Hornbostel-Sachs and its antecedents, including electrophones.



85

part of a working group for MIMO. So, there is no official ownership of Hornbostel-Sachs, other than the copyright of individual authors, translators and publishers. This places Hornbostel-Sachs in contrast with modern biblio-graphic schemes such as DDC, UDC and the second edi-tion of “Bliss Classification,” which have named editors, usually in conjunction with organisational structures of management and governance. Perhaps the closest Horn-bostel-Sachs has come to such an arrangement is through the MIMO consortium working group, chaired by Marga-ret Birley (Musical Instruments Museums Online 2011), which worked collectively on modifying Montagu’s revised version. Therefore, at least until the MIMO project, it could be argued that there was no official governance of Hornbostel-Sachs, and revision and maintenance was on an ad-hoc basis, usually determined by research interest ra-ther than through official review. 5.0 Reception of Hornbostel-Sachs The reception-related aspects of Hornbostel-Sachs are a significant part of this classification scheme. Lee’s (2015) reception studies framework is used to delineate the differ-ent strands of reception, to consider Hornbostel-Sachs’ consumption (usage), criticism and its Wirkung (the effects and influence of the scheme). The reception of Hornbos-tel-Sachs will be considered from the dual perspectives of its home domain (organology) and its impact on biblio-graphic classifications of music. 5.1 Usage of Hornbostel-Sachs Establishing the usage of classification schemes is meth-odologically difficult, as this information is rarely collected in a systematic fashion (Lee 2015). Discussions of con-sumption of Hornbostel-Sachs in organology discourse are mixed. On one hand, scholars describe Hornbostel-Sachs as a highly used classification scheme; for instance, it is labelled as “widely adopted” (Dournon 1992, 252) and “predominant” (Kolozali et al. 2011, 465), while Gnoli (2006) describes Hornbostel-Sachs as the most well-known and used of organological schemes. A more recent source, Weisser and Quanten (2011, 122) say that “[Horn-bostel-Sachs] is still used by most musical instruments mu-seums and in large inventory projects such as the MIMO,” and Koch and Kopal (2014) discuss its use at the Ethnol-ogisches Museum der Staatlichen Museen zu Berlin and other collections. On the other hand, Montagu and Burton (1971, 49) are not alone in stating that few organizations seem to use it to actually arrange their instruments. As little quantitative evidence is used to support either side of the argument, we are left to inferences. Kartomi (1990) provides a possible explanation of the differing views: while

many people use Hornbostel-Sachs, they are generally only using the first few steps or top few levels. This is illumi-nated by Ghirardini and Gnoli’s (2005) comments on the usage of Hornbostel-Sachs, as they imply that using only the first main classes of a scheme may be typical of a gen-eral pattern of how classification schemes are used. In other words, many people are using broad ideas from Hornbostel-Sachs, but there is little evidence from second-ary literature that many are using the full classification scheme. However, even with this cross-step it seems that there are conflicting accounts of Hornbostel-Sachs’ actual consumption when considering secondary accounts.

So, in the absence of quantitative primary data about us-age in more traditional settings of collections of instru-ments, broader examples of types of usage will be analysed instead to illuminate the different ways in which Hornbos-tel-Sachs is consumed. First, Hornbostel-Sachs is used in a number of different published resources about musical in-struments. It can be used to organise lists of instruments; for instance, Blades (1982) includes a checklist of percus-sion instruments from a particular collection, and this list is organised using Hornbostel-Sachs, including Hornbostel-Sachs notation and extensions. This type of usage could be considered equivalent to a classified catalogue of biblio-graphic items, such as the British Catalogue of Music (Coates 1960). Hornbostel-Sachs is also used in ethnomusi-cology and organology, as a way of organising a group of instruments found in the course of research. For instance, Picken (1977) uses Hornbostel-Sachs to organise a list of instruments found during field research in Afghanistan. In-terestingly, there is also evidence of Hornbostel-Sachs used as a pedagogical tool. In “Musical instruments of the world: an illustrated encyclopaedia” (1976), Hornbostel-Sachs is presented as a diagram in the encyclopaedia, as a represen-tation of the world of instruments. In addition, the basic categories of Hornbostel-Sachs are used to arrange the knowledge in this encyclopaedia, but note the encyclopaedia does not use Hornbostel-Sachs’ notation or order of classes. This use of organisation system to organise a textbook is an interesting example of KO in action, which perhaps could be related in part to Szostak’s (2018) development of KO systems as pedagogical tool for world history.

Second, it is interesting to consider Hornbostel-Sachs’ usage in the digital age. Perhaps the most significant exam-ple is the MIMO revision. This development of a new ver-sion of the classification scheme (see Section 4.4) was for a very specific use: an online museum of instruments, where the new version is specifically designed to work in a collaborative, digital environment. A second example of digital usage of Hornbostel-Sachs can be found in Wikipe-dia. For instance, broad categories such as aerophones have Wikipedia pages (List of Aerophones by Hornbostel-Sachs number 2019) which list the classes in these catego-



86

ries, the associated Hornbostel-Sachs notation and exam-ples of instruments in those classes (also sometimes add-ing extra examples not found in Hornbostel-Sachs itself). Such examples also make good use of hyperlinks to pro-vide what could be considered a digital, user-generated ver-sion of Hornbostel-Sachs. These lists also appear for the other three categories (List of chordophones by Hornbos-tel-Sachs number 2018, List of Idiophones by Hornbostel-Sachs number 2018, List of Membranophones by Horn-bostel-Sachs number 2018). (At the time of writing, the article for electrophones (Electrophone 2019) for electro-phone does not contain a classified list of instruments with corresponding notation, only information about this cate-gory.) Furthermore, some individual instruments have the “Hornbostel-Sachs classification” notation as a featured piece of metadata about the instrument in their Wikie-pedia entry, such as the flexatone (Flexatone 2018). These examples show how Hornbostel-Sachs has transcended being a way of organising physical objects or printed in-formation about instruments, and that Hornbostel-Sachs has been transformed for digital media and the digital age.

Finally, it is worth considering the use of Hornbostel-Sachs in the bibliographic sphere. Ghirardini and Gnoli (2005) discuss their survey of library use of Hornbostel-Sachs: they find it is little used. However, Ghirardini and Gnoli (2005) find that is Hornbostel-Sachs is used by one library, for its books about non-western music. This high-lights the comparatively cross-cultural nature of Hornbos-tel-Sachs compared to bibliographic classification schemes, and also could be seen as a sign that the scheme’s cross-cul-tural intentions are borne out in its reception. In addition, arguably Hornbostel-Sachs sees indirect usage in libraries through its influence on the contents and structure of bib-liographic schemes for music. This “Wirkung” is discussed in detail in Section 5.4. 5.2 Criticism of Hornbostel-Sachs Criticism of Hornbostel-Sachs is historically mixed. In pos-itive criticism, labels such as “monumental” (Grame 1963, 138) and “best” (Hood 1971, 125, describing a comment by Kunst) are used. Furthermore, examples of Hornbostel and Sachs in textbooks of music history such as Man’s Earliest Music (Carlin 1987) could be seen as acts of criticism; in other words, knowledge of Hornbostel-Sachs, a classifica-tion scheme, is seen as being crucial to knowledge of music history. Specific reasons for Hornbostel-Sachs’ perceived goodness are less common, but include its intended multi-cultural reach (Kartomi 2001). This shows how the cross-cultural intentions of Hornbostel-Sachs might have been re-alised in the execution of the scheme and appreciated by its audience. In discussing the use of Hornbostel-Sachs in museums, Koch and Kopal (2014, 300) mentions its “…klar

geregelten Charakteristika für die Ordnung von Instrumen-ten” (clear, regulated characteristics for the order of instru-ments), suggesting that its usage can be ascribed to its posi-tive, internal and structural qualities.

However, negative comments also abound, especially about specific aspects of the scheme. The inconsistency of how the four main classes (idiophones, chordophones, membranophones and aerophones) are subdivided is a noted disadvantage of the scheme; this can be seen in com-ments by Wachsmann (Wachsmann et al. 1980) and Kunst (described by Hood 1971). Other criticisms are noted, such as the confusing layout instigated by the decimal notation (Jairazbhoy 1990b) and the treatment of borderline instru-ments (Kartomi 1990). Furthermore, cross-classification is seen as an issue in Hornbostel-Sachs, as Kartomi (1990) also talks about issues with instruments that could live in two different places in the scheme (although this last criticism is suggested as a general problem with hierarchical classifica-tion rather than specifically with Hornbostel-Sachs). 5.3. Effect and influence of Hornbostel-Sachs The effect and influence of Hornbostel-Sachs (its “Wir-kung”) within organology can be seen in a number of ways. Examples already discussed in this article include the pres-ence of the introduction as an appendix to articles on the classification of instruments in various editions of “Grove” (Wachsmann et al. 1980, 1984, 2001), the new versions and translations of Hornbostel-Sachs in the twentieth and twenty-first centuries and the numerous new instrument classification schemes in the twentieth and twenty-first cen-turies that are adaptations of, or reactions to, Hornbostel-Sachs. The latter type of Wirkung includes new schemes created deliberately to attempt to resolve problems of Hornbostel-Sachs, such as Sakuri’s scheme, which explicitly demonstrates its disagreement to Hornbostel-Sachs by in-creasing the number of main categories (Dournon 1992). Lee (2014) has a longer discussion about these and other types of Wirkung of Hornbostel-Sachs.

Another way of viewing the effect of the original scheme is to reconceive it as a historical document which charts the development of discourse about instruments and culture. Koch and Kopal (2014, 301) discuss how the scheme deals with the “exotischer” (exotic) and “primitiv” (primitive), and suggest that the scheme is important to the study of scien-tific history. So, the qualities of the scheme that impinge Hornbostel-Sachs’ modern use, are the same aspects that also add to its influence, by way of aiding the study of the historical development of ethnomusicology, ethnology and organology.

Hornbostel-Sachs has also affected bibliographic classi-fication schemes for music, as seen by bibliographic classifications that have partially adopted Hornbostel-Sachs’s prin-



87

ciples, terminology and ideas. Conceptually, this means that a classification designed primarily for artefacts in the form of musical instruments is being transformed to classify mentefacts such as musical scores, and resources about mu-sical instruments.4 The scale of Hornbostel-Sachs’ infiltra-tion into bibliographic classification is noteworthy. Lee (2017b) found that out of a group of fifteen bibliographic classification schemes for music, seven made some use of Hornbostel-Sachs. In these examples, the level and types of influence varies; sometimes the bibliographic schemes bor-row Hornbostel-Sachs’s terminology, while sometimes they use its structure. Some examples of these influences are given in the next section; information about the methodol-ogy of these investigations, and differences between implicit and explicit factors, can be found in Lee (2017b). 5.4 Examples of Hornbostel-Sachs used in

bibliographic classification 5.4.1 Bliss Classification The first example is taken from the first edition of Bliss Classification (Bliss 1953). There is one Hornbostel-Sachs reference in this scheme, which comes in the VWT part of the schedules. This contains a class entitled “stringed instru-ments, chordophones;” so, the Hornbostel-Sachs category of chordophones is given as an alternative name for string instruments. There are no other signs of Hornbostel-Sachs usage in the arrangement or terminology of this scheme, showing how Hornbostel-Sachs can be used just as a pass-ing reference. 5.4.2 DDC early editions The second examples come from the pre-twentieth edi-tions of DDC; in other words, the examples are taken from editions before the revolutionary changes wrought by the “Phoenix Schedule” for music (Sweeney et al. 1980). The fifteenth edition of DDC, published in 1951, is the first edition of DDC to use Hornbostel-Sachs terminology: it chooses to describe as “membranophones” what had pre-viously been labelled “drum” (Dewey 1951), and this label is maintained until the major changes of the “Phoenix Schedule.” In addition, the fifteenth edition of DDC also uses the Hornbostel-Sachs term “electrophones;” how-ever, this term is dropped in the sixteenth to nineteenth editions. In light of the revolutionary nature of the fif-teenth edition of DDC, including its adoption of more modern terminology (Comaromi 1976), these examples of Hornbostel-Sachs terminology could be considered as a reflection on the perceived standing and symbolism of Hornbostel-Sachs. They position Hornbostel-Sachs as equating to “modern” in the eyes of DDC’s authors.

5.4.3 Flexible Classification The third example demonstrates a different type of Horn-bostel-Sachs usage: using Hornbostel-Sachs ideas to struc-ture part of the schedules. The “Flexible Classification” (Pethes 1967) uses Hornbostel-Sachs especially in the per-cussion part of the schedules.5 As well as using the Horn-bostel-Sachs terms “membranophones” and “idiophones,” these classes are also arranged in line with the Hornbostel-Sachs broad categories. Within each of the classes, various Hornbostel-Sachs ideas relating to how the sound is made (for instance, struck, shaken and friction) are used to organ-ize the classes, albeit the ideas are not employed in the same order as Hornbostel-Sachs. It is noteworthy that it is percus-sion that gains the detailed, explicit Hornbostel-Sachs make-over in “Flexible Classification.” Bibliographic schemes typ-ically treat percussion instruments as being less important than the other categories, which stems from the low im-portance attached to percussion within earlier periods of western art music (see Lee 2017b). So, we could see “Flexi-ble Classification” as a realisation in the bibliographic world of Hornbostel-Sachs’ cross-cultural aims. 5.4.4 UDC The fourth example is UDC, which has a fuller adoption of Hornbostel-Sachs. UDC is unusual in adopting Hornbostel-Sachs’s four main categories as its structure, and the scheme includes a plethora of Hornbostel-Sachs terminology and structural aspects. For example, like Hornbostel-Sachs, UDC has no keyboard category; instead, individual types of keyboard instrument are scattered amongst the main cate-gories. However, there are also ways that UDC does not fol-low Hornbostel-Sachs; for example, UDC’s classes are for individual instruments rather than characteristics of instru-ments, and UDC does not adopt Hornbostel-Sachs’ order within the idiophones category. Therefore, UDC shows how Hornbostel-Sachs can be a strong influence on struc-ture, terminology and order of concepts, without the bibli-ographic scheme entirely replicating Hornbostel-Sachs. 5.4.5 DDC Phoenix Schedule and modern editions The fifth example is the “Phoenix Schedule” of DDC, which is the basis for the DDC music schedules for the twentieh edition onwards. For instance, Hornbostel-Sachs terminology is used in the “Phoenix Schedule,” although it deliberately sits alongside more conventional names for in-struments and instrumental families; this again sets up the positioning of Hornbostel-Sachs as terminologically ad-vanced, with terms such as “strings” situated as the popular term. It is particularly insightful to consider the influence of Hornbostel-Sachs on the “Phoenix Schedule,” when read-



88

ing the scheme alongside the “Phoenix Schedule” authors’ stated intentions to base their scheme on Hornbostel-Sachs (Methodologically, this can be seen as using part of a multi-plane approach by delineating different types of infor-mation about classification schemes (Lee 2017a)). One ex-ample is keyboard instruments. These are given their own category in the “Phoenix Schedule,” despite this negating the fundamental feature of Hornbostel-Sachs, which has four broad categories separated by the method of sound production (for a fuller account, see Lee 2017b). Another example can be found in the inclusion and exclusion of clas-ses. While the authors of the “Phoenix Schedule” say they use Hornbostel-Sachs deliberately to create instrument schedules that are a “value-free basis for the classification” (Sweeney et al. 1980, xxii), so in other words will be less western-centric, actually many classes in Hornbostel-Sachs that do not have any western exemplars are missing from the DDC “Phoenix Schedule.” 6.0 Conclusion Hornbostel-Sachs is a highly significant classification within the theory and practice of organology. The scheme drew upon developments in instrument classification, such as Ma-hillon’s division into four categories, while purposefully de-signing a scheme that moves away from single-culture, west-ern-centric structures of instrument classification. Further-more, the scheme was highly original in its borrowing of the bibliographic idea of decimal classification, creating what the authors believed to be a culture-free notation as well as a way of representing the hierarchy of the scheme within its notation. While aiming to be universal in the cultures it cov-ered, the extensions and alternatives offered by Hornbostel-Sachs hint at strong localisation in the intended usage of the scheme.

Exploring the versions of Hornbostel-Sachs shows the establishment of the contents of the 1914 original as a mon-ument of music and organology history. This is evident from the translation of the original scheme nearly fifty years later after it was published and the inclusion of a translation of the 1914 scheme in a late twentieth century anthology of key texts in ethnomusicology. However, this article has shown that the representation of classification information by typographical means was not sacred, as this was generally not transmitted into translations or later versions.

An examination of the reception of Hornbostel-Sachs highlighted some interesting contrasts between perception and reality, especially in terms of its actual usage. While Hornbostel-Sachs seemed to be used for arranging collec-tions of instruments in the twentieth century, scholars such as Koch and Kopal (2014) have reservations about its suita-bility for this purpose. Furthermore, examining Hornbostel-Sachs’ reach into bibliographic classification, illustrates the

symbolism of Hornbostel-Sachs, showing how the scheme signifies technical knowledge and modernity. Perhaps the only true new “version” of the scheme is the MIMO ver-sion, although examining the germination of this version shows a complex and intriguing web of influences and rela-tionships. The MIMO version is particularly exciting for showing how a scheme from 1914 designed to organise physical collections of instruments, can be significant, dom-inating and versatile enough to be reimagined for a digital collection of instruments nearly a century later. Above all, Hornbostel-Sachs is shown to be a central classification scheme for curating and studying instruments, as well as playing a central role in musical instrument research and practice. Notes 1. Kartomi (1990) discusses how Galpin’s 1900 scheme

also utilises Mahillon’s four-category system, albeit with different names; however, as the next levels of hierar-chy within these categories did not follow Mahillon, and it is Mahillon that is mostly mentioned by Hornbostel and Sachs, Galpin’s 1900 scheme will not be discussed further.

2. According to Gnoli (2018), a decimal classification is usually associated with ten divisions, so there is a ques-tion about whether Hornbostel-Sachs’s notation can be called a “decimal notation” in the purest sense of the term.

3. Note that there is also, at the time of writing, an open access copy of this paper available from Montagu’s website dated from 2008, which is almost identical to the 2009 publication. For simplification, only the 2009 paper will be cited, as this is the published format.

4. Although, it should be noted that while primarily de-signed for the classification of instruments, Hornbos-tel-Sachs was also intended by its authors to be used for treatises and similar about instruments.

5. “A Flexible Classification System of Music and Litera-ture on Music” was written by the Hungarian music li-brarian, Ìvan Pethes, based on the UDC schedules and initiated by the International Association of Music Li-braries, Archives and Documentation Centres (IAML). The “Flexible Classification” aimed to be a universal classification for music literature and scores, which would bring together disparate classification practices into one scheme but appears to have resulted in little usage in libraries.

References Bakan, Michael B., Wanda Bryant, Guangming Li, David

Martinelli and Kathryn Vaughn. 1990, “Demystifying



89

and Classifying Electronic Music Instruments.” In Issues in Organology, ed. Sue Carole DeVale. Los Angeles: Uni-versity of California, Los Angeles, 37-64.

Barité, Mario. 2018. “Literary Warrant.” Knowledge Organi-zation 45: 517-36.

Beghtol, Claire. 2003. “Classification for Information Re-trieval and Classification for Knowledge Discovery: Re-lationships between “Professional” and “Naive” Classi-fications.” Knowledge Organization 30: 64-73.

Blades, James. 1982. A Checklist of the Percussion Instruments in the Edinburgh University Collection of Historic Musical In-struments. Edinburgh: Reid School of Music.

Bliss, Henry Evelyn. 1953. A Bibliographic Classification: Ex-tended by Systematic Auxiliary Schedules for Composite Specifi-cation and Notation. New York: H.W. Wilson.

British Standards Institution. 2006. UDC, Universal Decimal Classification. 3rd ed. London: British Standards Institu-tion.

Brown, Howard Mayer. 2001. "Instrument." Grove Music Online.

Carlin, Richard. 1987. Man’s Earliest Music. New York: Facts on File.

Coates, Eric. 1960. The British Catalogue of Music Classifica-tion. London: Council of the British National Bibliog-raphy.

Comaromi, John Phillip. 1976. The Eighteen Editions of the Dewey Decimal Classification. Albany, NY: Forest Press.

DeVale, Sue Carole. 1990. “Organizing Organology.” In Issues in Organology, ed. Sue Carole DeVale. Los Angeles: University of California, Los Angeles, 1-34.

Dewey Decimal Classification. 2016. EPC Exhibit 139-S78.1: 785.6-785: Ensembles Consisting of Only One Instru-mental Group. [Exhibit for Editorial Policy Committee 6-7 June 2016, exhibit by R. Green]. Dublin, OH: OCLC.

Dewey, Melvil. 1951. Decimal Classification. 15th ed. Milton James Ferguson, chairman and ed. Lake Placid, NY: Forest Press.

Diagram Group. 1976. Musical Instruments of the World: An Illustrated Encyclopaedia. London: Paddington.

“Electrophone.” 2019. Wikipedia. https://en.wikipedia. org/wiki/Electrophone

“Flexatone.” 2018. Wikipedia. https://en.wikipedia.org/ wiki/Flexatone

Dournon, Genevieve. 1992. "Organology." In Ethnomusi-cology: An Introduction, ed. Helen Myers. London: Mac-millan, 245-300.

Galpin, Francis W. 1937. A Textbook of European Musical Instruments. London: Williams & Norgate.

Ghirardini, Cristina and Claudio Gnoli. 2005. ´”Zampogne e Libri sulle Zampogne: Classificazioni Diverse.” Bibli-otime 8, no. 3. http://www.aib.it/aib/sezioni/emr/bib time/num-viii-3/gnoli.htm

Gnoli, Claudio. 2006. “Phylogenetic Classification.” Know- ledge Organization 33: 138-52.

Gnoli, Claudio. 2018a. "Mentefacts as a Missing Level in Theory of Information Science." Journal of Documentation 74: 1226-42. doi:10.1108/JD-04-2018-0054

Gnoli, Claudio. 2018b. “Notation.” Knowledge Organization 45: 667-84.

Gnoli, Claudio, Philippe Cousson, Tom Pullman, Gabriele Merli and Rick Szostak. 2011. “Representing the Struc-tural Elements of a Freely Faceted Classification.” In Classification & Ontology: Formal Approaches and Access to Knowledge: Proceedings of the International UDC Seminar, 19-20 September 2011, The Hague, The Netherlands, ed. Aida Slavic and Edgardo Civallero. Würzburg: Ergon, 193-205.

Grame, Theodore C. 1963. Review of Classification of Musical Instruments by Erich M. Von Hornbostel, Curt Sachs, An-thony Baines and Klaus P. Wachsmann. Ethnomusicology 7, no. 2: 137-8.

Guizzi, Febo. 2002. Guida alla Musica Popolare Italiana. 3: Gli instrumenti. Lucca: Libreria Musicale Italiana.

Heyde, Herbert. 1977. “Eine indische Klassifikation der Musikinstrumente: Bemerkungen zu den Systematiken von Bharata, Cassiodor, Hornbostel/Sachs, besonders Nārada.” Archiv für Musikwissenschaft 34, no. 2: 148-52.

Hood, Mantle. 1971. The Ethnomusicologist. New York: McGraw-Hill.

Hornbostel, Erich M. von and Curt Sachs. 1914. “Sys-tematik der Musikinstrumente. Ein Versuch.” Zeitschrift für Ethnologie: Organ der Berliner Gesellschaft für Anthropolo-gie, Ethnologie und Urgeschichte 46: 553-90.

Hornbostel, Erich M. von and Curt Sachs. 1992. “Classifi-cation of Musical Instruments.” In Ethnomusicology: An In-troduction, ed. Helen Myers. London: Macmillan, 444-61.

Hornbostel, Erich M. von and Curt Sachs. 1961. “Classifi-cation of Musical Instruments.” Trans. Anthony Baines and Klaus P. Wachsmann. Galpin Society Journal 14: 3-29. Translation of “Systematik der Musikinstrumente. Ein Versuch,” 1914. doi:10.2307/842168

“Hornbostel-Sachs.” 2019. Wikipedia. https://en.wikipe dia.org/wiki/Hornbostel%E2%80%93Sachs

Hjørland, Birger. 2008. “Classifying Madness: A Philo-sophical Examination of the Manual of Mental Disor-ders.” Knowledge Organization 35: 259-63.

Jairazbhoy, Nazir Ali. 1990a. “The Beginnings of Organ-ology and Ethnomusicology in the West: V. Mahillon, A. Ellis and S. M. Tagore.” In Issues in Organology, ed. Sue Carole DeVale. Los Angeles: University of California, Los Angeles, 67-80.

Jairazbhoy, Nazir Ali. 1990b. “An Explication of the Sachs-Hornbostel Instrument Classification System.” In Issues in Organology, ed. Sue Carole DeVale. Los Angeles: Uni-versity of California, Los Angeles, 81-104.



90

Kartomi, Margaret J. 1990. On Concepts and Classifications of Musical Instruments. Chicago: University of Chicago Press.

Kartomi, Margaret J. 2001. “The Classification of Musical Instruments: Changing Trends in Research from the Late Nineteenth Century, with Special Reference to the 1990s.” Ethnomusicology: Journal of the Society for Ethnomu-sicology 45: 283-314.

Katz, Israel J. 2001. “Hornbostel, Erich M(oritz) von.” Grove Music Online.

Kolozali, Sefki, Mathieu Barthet, Gyorgy Fazekas and Mark Sandler. 2011. “Knowledge Representation Issues in Musical Instrument Ontology Design.” In ISMIR 2011: Proceedings of the 12th International Society for Music Information Retrieval Conference, October 24–28, 2011, Mi-ami, Florida, ed. Anssi Klapuri and Colby Leider. Miami: University of Miami, 465-70.

Koch, Lars-Christian and Ricarda Kopal. 2014. “Klassi-fikation von Musikinstrumenten – Zum 100-jährigen Bestehen der Hornbostel-Sachs-Systematik.” Zeitschrift für Ethnologie 139: 281-302.

Lee, Deborah. 2014. “Webs of ‘Wirkung’: Modelling the In-terconnectedness of Classification Schemes.” In Know- ledge Organization in the 21st Century: Between Historical Pat-terns and Future Prospects: Proceedings of the Thirteenth Interna-tional ISKO conference, 19-22 May 2014, Krakow, Poland, ed. Wiesław Babik. Advances in Knowledge Organization 14. Würzburg: Ergon, 200-7.

Lee, Deborah. 2015. “Consumption, Criticism and Wir-kung: Reception-Infused Analysis of Classification Schemes.” Knowledge Organization 42: 508-21.

Lee, Deborah. 2017a. “Conceptions of Knowledge about Classification Schemes: A Multiplane Approach.” Infor-mation Research 22, no. 1. http://www.informationr.net/ ir/22-1/colis/colis1648.html

Lee, Deborah. 2017b. “Modelling Music: A Theoretical Ap-proach to the Classification of Notated Western Art Mu-sic.” PhD diss., City, University of London. http://open access.city.ac.uk/17445/

Lee, Deborah. 2017c. “Numbers, Instruments and Hands: The Impact of Faceted Analytical Theory on Classifying Music Ensembles.” Knowledge Organization 44: 405-15.

Lee, Deborah, Lyn Robinson and David Bawden. 2019. “Modelling the Relationship between Scientific and Bibliographic Classification for Musi.c” Journal of the Association of Information Science and Technology 70, no. 3: 230-41. doi:10.1002/asi.24120

“List of Aerophones by Hornbostel-Sachs number.” 2019. Wikipedia. https://en.wikipedia.org/wiki/List_of_aero phones_by_Hornbostel%E2%80%93Sachs_number

“List of Chordophones by Hornbostel-Sachs number.” 2018. Wikipedia. https://en.wikipedia.org/wiki/List_ of_chordophones_by_Hornbostel%E2%80%93Sachs

_number

“List of Idiophones by Hornbostel-Sachs number.” 2018. Wikipedia. https://en.wikipedia.org/wiki/List_of_idio phones_by_Hornbostel%E2%80%93Sachs_number

“List of Membranophones by Hornbostel-Sachs number.” 2018. Wikipedia. https://en.wikipedia.org/wiki/List_ of_membranophones_by_Hornbostel%E2%80%93

Sachs_number Mai, Jens-Erik. 2011. “The Modernity of Classification.”

Journal of Documentation 67, no. 4: 710-730. doi:10.1108/ 00220411111145061

Mahillon, Victor-Charles. 1880. Catalogue Descriptif & Ana-lytique du Musée Instrumental du Conservatoire Royal de Brux-elles: Précédé d’un Essai de Classification de Tous les Instruments Anciens et Modernes. Ghent: C. Annoot-Braeckman.

Montagu, Jeremy and John Burton. 1971. “A Proposed New Classification System for Musical Instruments.” Ethnomusicology 15: 49-70.

Montagu, Jeremy. 2009. “It’s Time to Look at Hornbostel-Sachs Again.” Muzyka: Kwartalnik Poświęcony Historii i Teo- rii Muzyki 54, no. 1: 7-28.

Montagu, Jeremy. 2008. “It’s Time to Look at Hornbostel-Sachs again: Hornbostel-Sachs Reconsidered.” http:// www.jeremymontagu.co.uk/Hornbostel-Sachs%20Re

considered.pdf MIMO (Musical Instruments Museums Online). 2011. “Re-

vision of the Hornbostel-Sachs Classification of Musical Instruments by the MIMO Consortium.” Last modified 8 July. http://network.icom.museum/fileadmin/user_ upload/minisites/cimcim/documents/H-S_20classifica

tion_20final_20version_20_282013_29_20without_20 editorial_20markings-2.pdf Olsen, Dale. 1986. “"Is it Time for Another - Phone?” SEM

newsletter 20, no. 4: 5. Pethes, Ìvan. 1967. A Flexible Classification System of Music

and Literature on Music. Preprint ed. Budapest: Centre of Library Science and Technology.

Picken, Laurence E. R. “Some Children’s Sound-Producing Toys and Other ‘Primitive’ Musical Instruments from Af-ghanistan.” In Neue Ethnomusikologische Forschungen: Fest-schrift Felix Hoerburger zum 60. Geburtstag am 9. Dezember 1976, ed. Peter Baumann, Rudolf Maria Brandl and Kurt Reinhard. Regensburg: Laaber-Verlag, 177-89.

Ramey, Michael. 1974. “A Classification of Musical Instru-ments for Comparative Study.” PhD diss., University of California Los Angeles.

Sachs, Curt. 1914. “Die Hornbostel-Sachs’sche Klassifika-tion der Musikinstrumente.” Die Naturwissenschaften. 2, no. 51: 1056-9

Sachs, Curt. 1940. The History of Musical Instruments. New York: W.W. Norton.

Sweeney, Russell, John Clews and Winton E. Matthews. 1980. DDC: Dewey Decimal Classification: Proposed Revision



91

of 780 Music Based on Dewey Decimal Classification and Rel-ative Index. Albany, NY: Forest Press.

Szostak, Rick. 2014. “How Universal is Universality?” Know- ledge Organization 41: 468-70.

Szostak, Rick. 2018. “Applied Knowledge Organization and the History of the World.” In Challenges and Opportunities for Knowledge Organization in the Digital Age: Proceedings of the Fifteenth International ISKO Conference, 9-11 July 2018, Porto, Portugal, ed. Fernanda Ribeiro and Maria Elisa Cerveira. Advances in Knowledge Organization 16. Würzburg: Er-gon, 52-7.

Tennis, Joseph. T. 2010, “Measured Time: Imposing a Tem-poral Metric to Classificatory Structures.” In Paradigms and Conceptual Systems in Knowledge Organization: Proceedings of the Eleventh International ISKO Conference 23-26 February 2010, Rome, Italy, ed. Claudio Gnoli and Fulvio Maz-zocchi. Advances in Knowledge Organization 12. Würz-burg: Ergon, 223-8.

Wachsmann, Klaus, Erich M. von Hornbostel and Curt Sachs. 1984. “Classification.” In The New Grove Dictionary of Musical Instruments, ed. Stanley Sadie. London: Macmil-lan, 407-10.

Wachsmann, Klaus, Erich M. von Hornbostel and Curt Sachs. 1980. “Instruments, Classification of.” In The New Grove Dictionary of Music and Musicians, ed. Stanley Sadie. London: Macmillan, 237-45.

Wachsmann, Klaus, Kartomi, Margaret, Hornbostel, Erich M. von and Sachs, Curt. 2001. “Instruments, Classifica-tion of.” Grove Music Online.

Weisser, Stéphanie and Maarten Quanten. 2011. “Rethink-ing Musical Instrument Classification: Towards a Mod-ular Approach to the Hornbostel-Sachs System.” Year-book for Traditional Music 43: 122-46.


Letters to the Editor

92

Letter to the Editor

DOI:10.5771/0943-7444-2020-1-92

Popularity of entries in ISKO Encyclopedia of Know-ledge Organization The ISKO Encyclopedia of Knowledge Organization (IEKO) was launched in 2016 by Birger Hjørland, its Editor-in-chief, as an official ISKO initiative; Claudio Gnoli joined soon as co-editor and web editor. Peer-reviewed articles are publi-shed online at http://www.isko.org/cyclo/ then printed in the Knowledge Organization journal (Dextre Clarke 2017).

Since 2018, the Web version of new entries includes a counter of independent visits provided by Digits.net; the counter has also been progressively introduced for all previ-ously-published entries, keeping track of the date when the count has started. After a couple of years, such statistics of-fer an interesting hint to assess which topics are the most popular in our field. Obviously, this is not an objective mea-sure of the absolute relevance of a topic or quality of a page: for example, an entry on a very specific topic can be expec-ted to be consulted less often than those on more general topics, yet still be a necessary component in the documen-tation of knowledge organization (KO) concepts.

On 8 November 2019, we have tabulated the current value of counters for 46 IEKO entries. The other 11 ent-ries available at that time have not been considered, as they still had not had a counter for a period significant enough (at least 40 days). Visits for an individual page ranged between 113 and 9010. As these values are clearly biased by the different age of each counter, we have weighed them by the number of days elapsed since the introduction of the counter (often, though not always, coinciding with the entry creation). Number of elapsed days ranged between 44 and 604.

Dividing the former value by the latter, we got a visit rate v for every entry. Resulting values of v range between 0.89 and 17.36 visits per day per entry, with a mean of 4.11. The ten most often visited entries are as shown in Table 1.

There are many possible ways to explain these results. A first observation is that the most visited entries concern very general topics in KO and the broader field of library-and-information science (LIS)—as opposed to, for exa-mple, knowledge organization systems (KOSs) in specific fields or biographical articles on individual KO authors. This may reflect a use of IEKO in educational contexts, contributing to a greater awareness of the basics of our field among non-specialists.

Exceptions to this are the entries on Hornbostel-Sachs and on the classification of psychology, which may have been largely used due to the popularity of the subject as taught in specific KO courses or to the renown of their authors. In general, humanities may be of greater interest to the KO community than other covered fields, such as physics or astronomy, although this hypothesis would need further evidence.

The systematic index of IEKO is organized by broad categories that are identified by capital letters (compatible with the Integrative Levels Classification (ILC) notational system for special and local schemes) and used in anchor links. We have aggregated data on visit rates by such cate-gories and calculated the average v for each category and subcategory. Results are shown in Table 2.

As can be seen, general entries on the discipline itself (entry on “KO”) and adjacent disciplines (entry on “LIS”) have by far the highest average v, confirming that users’ interests focus on introductory resources. Apart from this,

17.36 Knowledge pyramid: the DIKW hierarchy

14.83 Library and information science (LIS)

11.60 Knowledge organization (KO)

11.49 Classification

6.92 Hornbostel-Sachs Classification of Musical Instruments

6.91 Literary warrant

6.58 Citation indexing and indexes

6.27 Knowledge organization system (KOS)

6.17 Indexing: concepts and theory

6.13 Classification of psychology

Table 1.



93

the average values for all broad categories do not differ very much. The low value for general KOSs can be explai-ned by the fact that entries for the most renowned systems (DDC, UDC, BC2 ...) are still in preparation or (in the case of Colon Classification) have lacked a counter until recently so are not included in this survey. Claudio Gnoli and Edoardo Manelli Library Service, University of Pavia

Via Ferrata 1, 27100 Pavia, Italy <[email protected]> <[email protected]> Reference Dextre Clarke, Stella. 2017. “Babel, Babble and Info-speak:

Could an Encyclopedia Help?”. Catalogue and Index 188: 59-60.

Letter to the Editor

DOI:10.5771/0943-7444-2020-1-93

Databases should Keep Pace with the Needs of sci-entific Exploration: “Nationality” should be added to scientific Research Databases The rapid development of science and technology has shor-tened the distances among people from different countries and regions. Many people study or work abroad rather than in their home countries. According to Decoding Global Ta-lent 2018 (https://on.bcg.com/2tB3qy7), 57% of respon-dents expressed willingness to work abroad. Working abroad has become a global trend. At the same time, research

on countries or regions has always been a hot topic. A large number of results can be obtained when searching for a country, a region, developing country, or developed country in Google Scholar. The question arises: How do we consider the impact of those who work abroad on related research?

It is difficult to assess the specific impact of talents on national development and social progress. Even the most intuitive literature analysis work is also facing difficulties. A great deal of literature analysis is based on Science Citation Index and Social Sciences Citation Index in the Web of Science database. However, it should be noted that the “Count-

5.53 A KO: general and historical issues 13.21 AD Discipline and adjacent disciplines

1.68 AR Biographical articles

5.35 C Core concepts in KO 5.46 CC Theoretical concepts

4.16 CS Specific document types, genres and media

3.29 K Knowledge organization systems (KOS) 4.98 KA KOS general issues

5.21 KD KOS kinds

1.87 KG Specific KOSs, general/universal

2.85 KL Specific KOSs, domain/specific

2.90 KN KO in specific domains

2.87 KS Standards and formats for representing data

4.83 P Knowledge organizing processes (KOP) 2.48 R Methods, approaches and philosophies 2.09 T KO in different contexts and applications

Table 2.



94

ries/Regions Search” in the Web of Science database refers to the countries in which the authors work. There is no relevant nationality information in the Web of Science data-base. Nationality information is crucial to the rigor and ac-curacy of relevant research. Do researchers manually coll-ect nationalities of so many unfamiliar people one by one? It sounds impractical and absurd.

Faced with such a problem, it is particularly necessary to add nationality information to the scientific research database, and it can bring the following benefits for future research: 1. Save time for relevant research staff. 2. Manual processing in a study is often difficult to verify.

If there are nationality-related items in the database, re-viewers or readers can easily and accurately verify the findings when they have doubts.

3. The number of highly skilled talents working abroad and their institutions in any field can be accurately and quickly obtained. That is to say, it can help us track spe-cific data about talents flow in any field of any country. The accurate data on the changes in people working ab-road will be easily obtained. The information about ta-lent flow obtained in such a way is certainly more accu-rate and helpful than the sample interview. What’s more, the cost of research will be reduced compared to the troublesome interview survey. And surveys like De-coding Global Talent 2018 will be easier and more con-vincing, and perhaps the findings will be more valuable than existing research.

4. When there is enough data about nationality in the fu-ture, it enables us to carry out some interesting research, for example, comparing the number of achievements, research directions, and other valuable aspects between native and foreign talents in any field of any country. In

addition, a series of studies can be carried out and com-pared with existing studies to better understand social problems and promote global progress.

Therefore, I suggest that databases like Web of Science should include nationality-related items. It is not my o-pinion to determine the author's nationality one by one for those articles that have been published. I do not want to bring trouble to the staff of databases and publishers. Rather, I suggest that the newly published studies contain “nationality” from a certain time in the future. And 3 years, 5 years or 10 years later, there will be enough samples for scholars to carry out a series of studies.

Finally, it must be noted that not just the Web of Science database that needs to be improved and not just the “nati-onality” problem that needs to be solved. I hope to attract more innovative databases or other scientific research tools through the “nationality” problem. With the pro-gress of the times, if the indicators in the database remain unchanged, they may not be able to keep up with the needs of scientific exploration. A little change today maybe pro-vides valuable contributions for future research. Why don't we do that? Guangyue Wei Xidian University Xian, China <[email protected]> References Strack, Rainer, Mike Booker, Orsolya Kovács-Ondrejko-

vic, Pierre Antebi and David Welch. 2018. “Decoding Global Talent 2018” (blog). Atlanta; Boston: Boston Consulting Group. https://on.bcg.com/2tB3qy7.


Books Recently Published

95

Books Recently Published Compiled by J. Bradford Young

DOI:10.5771/0943-7444-2020-1-95

Asokan, G. 2020. Science of Language in Vedic Tradition: Pho-

netics, Grammar, Logic, & Epistemology. Delhi: Parima. Ballestro, John, ed. 2020. The Library’s Guide to Graphic Nov-

els. Chicago: ALA Editions. Baumann, Benjamin and Daniel Bultmann, eds. 2020. So-

cial Ontology, Sociocultures and Inequality in the Global South. New York: Routledge.

Bursten, Julia R.S. 2020. Perspectives on Classification in Syn-thetic Sciences: Unnatural Kinds. New York: Routledge.

Chimakonam, Jonathan O. and Uchenna L. Ogbonnaya, eds. 2020. African Metaphysics, Epistemology and a New Logic: Emerging Research and Opportunities. Hershey, PA: Information Science Reference.

Crowder, Jerome W., Mike Fortun, Rachel Besara, Lindsay Poirier, eds. 2020. Anthropological Data in the Digital Age: New Possibilities – New Challenges. Cham, Switzerland: Palgrave Macmillan.

Das, Asit Kumar, Janmenjoy Nayak, Bighnaraj Naik, Soumen Kumar Pati, Danilo Pelusi, eds. 2020. Computa-tional Intelligence in Pattern Recognition: Proceedings of CIPR 2019. Singapore: Springer.

Fiori, Alessandro. 2020. Trends and Applications of Text Sum-marization Techniques. Hershey, PA: IGI Global.

Fricker, Miranda, Peter J. Graham, David Henderson and Nikolaj J. L. L. Pedersen, eds. 2020. The Routledge Hand-book of Social Epistemology. New York: Routledge.

Hirsh, Sandra and Susan Alman. 2020. Blockchain. Chicago: ALA Neal-Schuman.

Jal, Murzban and Jyoti Bawane, eds. 2020. Theory and Praxis: Reflections on the Colonization of Knowledge. New York: Routledge.

Jennex, Murray Eugene, ed. 2020. Current Issues and Trends in Knowledge Management, Discovery and Transfer. Hershey, PA: Information Science Reference.

Ji, Meng. 2020. Translating and Communicating Environmental Cultures. Routledge Studies in Empirical Translation and Multilingual Communication. New York: Routledge.

Katerere, David R., Wendy Applequist, Oluwaseyi M. Aboyade and Chamunorwa Togo, eds. 2020. Traditional and Indigenous Knowledge for the Modern Era: A Natural and Applied Science Perspective. Boca Raton, FL: CRC Press.

McCain, Kevin and Kostas Kampourakis. 2020. What Is Scientific Knowledge? An Introduction to Contemporary Episte-mology of Science. New York: Routledge.

Mendoza-González, Ricardo, Huizilopoztli Luna-García and Alfredo Mendoza-González, eds. 2020. UXD and UCD Approaches for Accessible Education. Hershey, PA: In-formation Science Reference.

Olson, David L. and Desheng Wu. 2020. Predictive Data Mining Models. 2nd ed. Singapore: Springer.

Pernecky, Tomas, ed. Postdisciplinary Knowledge. New York: Routledge.

Wilson, Emma Annette, ed. 2020. Digital Humanities for Li-brarians. Lanham, MD: Rowman & Littlefield.


Index to Volume 46 (2019)

96

Index to Volume 46 (2019) No. 1, pp. 1-74; No. 2, pp. 75-155; No. 3, pp. 159-240; No. 4, pp. 243-333;

No. 5, pp. 335-400; No. 6, pp. 401-489; No. 7, pp. 491-569; No. 8, pp. 571-669.

DOI:10.5771/0943-7444-2020-1-96

ALPHABETICAL INDEX 1. Articles Bak, Greg, Danielle Allard, and Shawna Ferris. Knowledge Organization as Knowledge Creation: Surfacing Community Participation in Archival Arrangement and Description ..................................................... 502 Barité, Mario. Towards a General Conception of Warrants: First Notes .................................................................... 647 Bragato Barros, Thiago Henrique and Renato Tarciso Barbosa de Sousa. Archival Science and Knowledge Organization: Mapping Methodological Relationships .................................................................................. 493 Broughton, Vanda. The Respective Roles of Intellectual Creativity and Automation in Representing Diversity: Human and Machine Generated Bias ....................... 596 Chen, Lielei and Hui Fang. An Automatic Method for Extracting Innovative Ideas Based on the Scopus® Database .......................................................................................... 171 Chen, Shu-Jiun. Semantic Enrichment of Linked Archival Materials .......................................................................... 530 Chen, Shu-Jiun. Semantic Enrichment of Linked Personal Authority Data: A Case Study of Elites in Late Imperial China ....................................................................... 607 Clarke, Stella G. Dextre. The Information Retrieval Thesaurus........................................................................................ 439 Clavier, Viviane. Knowledge Organization, Data and Algorithms: The New Era of Visual Representations ............. 615 Coen, Gerard and Richard P. Smiraglia. Toward Better Interoperability of the NARCIS Classification ......................... 345 de Mattos Souza, Rosale. The Representation of Archival Information in Controlled Vocabularies: The Context of the Archival Institutions in Rio de Janeiro ............................................................................................ 548 Dick, Steven J. Astronomy’s Three Kingdom System: A Comprehensive Classification System of Celestial Objects ............................................................................................ 460 Finnemann, Niels Ole. Web Archive ............................................ 47

Freeman, Graham and Robert J. Glushko. Organization, Not Inspiration: A Historical Perspective of Musical Information Architecture ......................................... 161 Frické, Martin. The Knowledge Pyramid: the DIKW Hierarchy .......................................................................................... 33 Gomes, Pablo and Maria Guiomar da Cunha Frota. Knowledge Organization from a Social Perspective: Thesauri and the Commitment to Cultural Diversity .............. 639 Hauser, Elliott and Joseph T. Tennis. Episemantics: Aboutness as Aroundness ............................................................ 590 Johansson, Sandra and Koraljka Golub. LibraryThing for Libraries: How Tag Moderation and Size Limitations Affect Tag Clouds ......................................................................... 245 Koraljka Golub. Automatic Subject Indexing of Text ............ 104 Korwin, Wendy and Haakon Lund. Alphabetization ............... 209 Kwaśnik, Barbara H. Changing Perspectives on Classification as a Knowledge-Representation Process ........... 656 Legendere, Ariadne. The Development of the Canadian Research and Development Classification ............... 371 Lin, Feng-Tyan. Drawing a Knowledge Map of Smart City Knowledge in Academia ........................................... 419 Lu, Wei, Xin Li, Zhifeng Liu and Qikai Cheng. How do Author-Selected Keywords Function Semantically in Scientific Manuscripts? ............................................................. 403 Moeller, Robin A. and Kim E. Becnel. “Why on Earth Would We Not Genrefy the Books?”: A Study of Reader-Interest Classification in School Libraries .................... 199 Naskar, Debashis and Subhashis Das. HNS Ontology Using Faceted Approach .............................................................. 187 Nikiforova, Aleksandra A. Soil Classification ........................... 467 Quinlan, Emma and Pauline Rafferty. Astronomy Classification: Towards a Faceted Classification Scheme ........ 260 Rafferty, Pauline. Disrupting the Metanarrative: A Little History of Image Indexing and Retrieval ....................... 4 Saarti, Jarmo. Fictional Literature, Classification and Indexing .................................................................................. 320


Index to Volume 46 (2019)

97

Sandy, Heather Moulaison and Andrew Dillon. Mapping the KO Community ...................................................... 578 Schöpfel, Joachim, Dominic Farace, Hélène Prost and Antonella Zane. Data Papers as a New Form of Knowledge Organization in the Field of Research Data ................................................................................................. 622 Smiraglia, Richard P. Trajectories for Research: Fathoming the Promise of the NARCIS Classification ........... 337 Smiraglia, Richard P. Work ........................................................... 308 Stuart, Emma. Flickr: Organizing and Tagging Images Online ................................................................................ 223 Tognoli, Natalia and José Augusto Chaves Guimarães. Provenance as a Knowledge Organization Principle ............... 558 Tognoli, Natália Bolfarini, Ana Célia Rodrigues, and José Augusto Chaves Guimarães Archival Knowledge: Conceptual Frameworks for Recent Terminology in the KO Domain ................................................................................... 522 Vancauwenbergh, Sadia and Hanne Poelmans. The Flemish Research Discipline Classification Standard: A Practical Approach .................................................................... 354 Wang, Shenghui and Rob Koopman. Embed First, Then Predict ................................................................................... 364 White, Howard D. Patrick Wilson ............................................... 279 Xu, Liwei and Jiangnan Qiu. Unsupervised Multi-class Sentiment Classification Approach ............................................... 15 Yang, Li and Yejun Wu. Creating a Taxonomy of Earthquake Disaster Response and Recovery for Online Earthquake Information Management ........................... 77 Zeng, Marcia Lei. Interoperability ............................................... 122 Zhang, Chengzhi, Hua Zhao, Xuehua Chi and Shuitian Ma. Information Organization Patterns from Online Users in a Social Network ....................................... 90

2. Brief Communications Solc, Roman. Brief Communication: Some Critical Notes on the Czech System of Evaluation of Research .......................................................................................... 387 Thorat, Rahul and Reinout van Brakel. Brief Communication: The Need for a National-Level Working Group for Higher Education Research Data in The Netherlands ....................................................................... 380 3. Features, Editorials, Etc. Annual Progress in Thesaurus Research? .................................. 238 Broughton, Vanda. Emeritus Professor Ia McIlwaine: An Appreciation ............................................................................ 573 Peer Review in 2018 .......................................................................... 3 Sixth Annual “Best Article in KO Award” ................................ 668 Xiao, Guohua. Explainable AI: Linking Human and Machine ................................................................................... 398 Xiao, Guohua. The Brain is a Knowledge Graph ...................... 71 4. Reviews Semantic Perception: How the Illusion of a Common Language Arises and Persists, Jody Azzouni. New York: Oxford University Press, 2013. ISBN 9780199967407. 2015. ISBN 9780190275549. Ontology Without Borders, Jody Azzouni. New York, NY: Oxford University Press. ISBN 9780190622558. Reviewed by Matthew Kelly .......................................................................... 147 5. Books Recently Published .......... 72, 155, 240, 333, 400, 489, 569, 669 6. Other Erratum ........................................................................................... 154 Index to Volume 45 ........................................................................ 73


KNOWLEDGE ORGANIZATION KO Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444


Publisher Ergon – ein Verlag in der Nomos Verlagsgesellschaft mbH Waldseestraße 3-5 D-76530 Baden-Baden Tel. +49 (0)7221-21 04-667 Fax +49 (0)7221-21 04-27 Sparkasse Baden-Baden Gaggenau IBAN: DE05 6625 0030 0005 0022 66 BIC: SOLADES1BAD KNOWLEDGE ORGANIZATION Journal of the International Society for Knowledge Organization Richard P. Smiraglia, Editor-in-Chief [email protected] Instructions for Authors Manuscripts should be submitted electronically (in Microsoft® Word format) in English only via ScholarOne at https://mc04.manuscriptcentral.com/jisko. Manuscripts that do not adhere to these guidelines will be returned to the authors for resubmission in proper form.

Manuscripts should be accompanied by an indicative abstract of approximately 250 words. Manuscripts of articles should fall within the range 6,000-10,000 words. Longer man-uscripts will be considered on consultation with the Editor-in-chief.

A separate title page should include the article title and the author’s name, postal ad-dress, and E-mail address. Only the title of the article should appear on the first page of the text. Contact information must be present for all authors of a manuscript.

To protect anonymity, the author’s name should not appear on the manuscript. Criteria for acceptance will be appropriateness to the field of knowledge organization

(see “Scope” at https://www.isko.org/ko.html), taking into account the merit of the contents and presentation. It is expected that all successful manuscripts will be well-situ-ated in the domain of knowledge organization, and will cite all relevant literature from within the domain. Authors are encouraged to use the KO literature database at http://www.isko.org/lit.html.

The manuscript should be concise and should conform to professional standards of English usage and grammar. Authors whose native language is not English are encouraged to make use of professional academic English-language proofreading services. We recom-mend Vulpine Academic Services ([email protected]).

Manuscripts are received with the understanding that they have not been previously published, are not being submitted for publication elsewhere, and that if the work received official sponsorship, it has been duly released for publication. Submissions are refereed, and authors will usually be notified within 6 to 8 weeks. Manuscript Formatting Under no circumstances should the author attempt to mimic the presentation of text as it appears in our published journal. Instead, please follow these instructions:

In Microsoft® Word please set the language preference (“Tools,” “Language”) to “Eng-lish (US)” or “English (UK).”

The entire manuscript should be double-spaced, including notes and references. The text should be structured with decimally-numbered subheadings (1.0, 1.1, 2.0, 2.1,

2.1.1, etc.). It should contain an introduction, giving an overview and stating the purpose, a main body, describing in sufficient detail the materials or methods used and the results or systems developed, and a conclusion or summary.

Author-generated keywords are not permitted. Footnotes are not allowed but endnotes may be used. Do not use automatic footnote

formatting. Instead, insert a superscript numeral (Format, Font, Superscript) and create the text of the note manually in a separate list at the end of the manuscript, before the reference list.

Paragraphs should include a topic sentence, a developed narrative and a conclusion; a typical paragraph has several sentences. Paragraphs with tweet-like characteristics (one or two sentences) are inappropriate.

Italics are permitted only for phrases from languages other than English, and for the titles of published works.

Bold type is not permitted. Em-dashes should not be used as substitutes for commas. Dashes must be inserted

manually (Insert, Advanced Symbol, Em-dash) with no spaces on either side. Illustrations should be embedded within the document. Photographs (including color

and half-tone) should be scanned with a minimum resolution of 600 dpi and saved as .jpg files. Tables should contain a number and caption at the bottom, and all columns and rows should have headings. All illustrations should be cited in the text as Figure 1, Figure 2, etc. or Table 1, Table 2, etc.

Examples of KOS (classification, taxonomy, ontology, thesaurus, etc.) arrays should be configured as figures and set into the document as jpgs; they should not be entered as ed-itable text.

Remove all active hyperlinks, including those from reference formatting software (if hovering over the text with a mouse produces a gray highlight, the text is hyperlinked; re-move the link “Insert,” “Hyperlink,” “Remove link”).

Upon acceptance of a manuscript for publication, authors must provide a digital photo and a one-paragraph biographical sketch (fewer than 100 words). The photograph should be scanned with a minimum resolution of 600 dpi and saved as a .jpg file. References Reference citations within the text should have the form: (Author year). For example, (Jones 1990). Specific page numbers are required for quoted material, e.g. (Jones 1990, 100). A citation with two authors would read (Jones and Smith 1990); three or more authors would be: (Jones et al. 1990). When the author is mentioned in the text, only the date and optional page number should appear in parentheses: “According to Jones (1990), …” or “Smith wrote (2010, 146): ….” A subsequent page reference to the same cited work (e.g., to Smith 2010) should have the form “(229).” There is never a comma before the date.

In-text citations should not be routinely placed at the end of a sentence or after a quo-tation, but an attempt should be made to work them into the narrative. For example:

Jones (2010, 114) reported statistically significant results. Many authors report similar data; according to Matthews (2014, 94): “all seven studies

report means within ±5%.” In-text citations should precede block quotations, and never are placed at the end of a

block-quotation. References should be listed alphabetically by author at the end of the article. Reference

lists should not contain references to works not cited in the text. Websites mentioned in passing in the text should be identified parenthetically with their URLs but not with refer-ences unless a specific page of a specific website is being quoted.

Author names should be given as found in the sources (not abbreviated, but also not fuller than what is given in the source). Journal titles should not be abbreviated. Multiple citations to works by the same author should be listed chronologically and should each in-clude the author’s name. Articles appearing in the same year should have the following for-mat: “Jones 2005a, Jones 2005b, etc.”

Proceedings must be identified fully by volume title (which often differs from the name of the conference), editor(s), series (if applicable), and details of publication (place and publisher name). Papers in proceedings must be identified by their page numbers. Papers from online proceedings that lack publication details should include a DOI or URL.

Journal issue numbers are given only when a journal volume is not through-paginated. References for published electronic resources should be accompanied by either a URL or DOI but not in lieu of actual publication data; access dates are not allowed.

Unpublished electronic resources may use an access date in lieu of a data of publica-tion. In cases of doubt, authors are encouraged to consult The Chicago Manual of Style 17th ed. (or online), author-date reference system (chapter 15).

Examples: Dahlberg, Ingetraut. 1978. “A Referent-Oriented, Analytical Concept Theory for INTER-

CONCEPT.” International Classification 5: 142-51. Howarth, Lynne C. 2003. “Designing a Common Namespace for Searching Metadata-Ena-

bled Knowledge Repositories: An International Perspective.” Cataloging & Classification Quarterly 37, nos. 1/2: 173-85.

Pogorelec, Andrej and Alenka Šauperl. 2006. “The Alternative Model of Classification of Belles-Lettres in Libraries.” Knowledge Organization 33: 204-14.

Schallier, Wouter. 2004. “On the Razor’s Edge: Between Local and Overall Needs in Knowledge Organization.” In Knowledge Organization and the Global Information Society: Pro-ceedings of the Eighth International ISKO Conference 13-16 July 2004 London, UK, edited by Ia C. McIlwaine. Advances in knowledge organization 9. Würzburg: Ergon Verlag, 269-74.

Smiraglia, Richard P. 2001. The Nature of ‘a Work’: Implications for the Organization of Knowledge. Lanham, Md.: Scarecrow.

Smiraglia, Richard P. 2005. “Instantiation: Toward a Theory.” In Data, Information, and Knowledge in a Networked World; Annual Conference of the Canadian Association for Information Science … London, Ontario, June 2-4 2005, ed. Liwen Vaughan. http://www.cais-acsi.ca/2005proceedings.htm.

© Ergon – ein Verlag in der Nomos Verlagsgesellschaft, Baden-Baden 2020. All Rights reserved. KO is published by Ergon. Annual subscription 2020: – Print + online (8 issues/ann.; unlimited access for your Campus via Nomos

eLibrary) € 359,00/ann. – Prices do not include postage and packing – Cancellation policy: Termination within 3 months‘ notice to the end of the cal-

endar year


KO KNOWLEDGE ORGANIZATION Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444


Scope

The more scientific data is generated in the impetuous present times, the

more ordering energy needs to be expended to control these data in a

retrievable fashion. With the abundance of knowledge now available the

questions of new solutions to the ordering problem and thus of im-

proved classification systems, methods and procedures have acquired un-

foreseen significance. For many years now they have been the focus of

interest of information scientists the world over.

Until recently, the special literature relevant to classification was pub-

lished in piecemeal fashion, scattered over the numerous technical jour-

nals serving the experts of the various fields such as:

philosophy and science of science

science policy and science organization

mathematics, statistics and computer science

library and information science

archivistics and museology

journalism and communication science

industrial products and commodity science

terminology, lexicography and linguistics

Beginning in 1974, KNOWLEDGE ORGANIZATION (formerly IN-

TERNATIONAL CLASSIFICATION) has been serving as a common

platform for the discussion of both theoretical background questions

and practical application problems in many areas of concern. In each is-

sue experts from many countries comment on questions of an adequate

structuring and construction of ordering systems and on the problems

of their use in opening the information contents of new literature, of

data collections and survey, of tabular works and of other objects of sci-

entific interest. Their contributions have been concerned with

(1) clarifying the theoretical foundations (general ordering theory/

science, theoretical bases of classification, data analysis and reduc-

tion)

(2) describing practical operations connected with indexing/classifi-

cation, as well as applications of classification systems and the-

sauri, manual and machine indexing

(3) tracing the history of classification knowledge and methodology

(4) discussing questions of education and training in classification

(5) concerning themselves with the problems of terminology in gen-

eral and with respect to special fields.

Aims Thus, KNOWLEDGE ORGANIZATION is a forum for all those in-terested in the organization of knowledge on a universal or a domain-specific scale, using concept-analytical or concept-synthetical approaches, as well as quantitative and qualitative methodologies. KNOWLEDGE ORGANIZATION also addresses the intellectual and automatic compi-lation and use of classification systems and thesauri in all fields of knowledge, with special attention being given to the problems of termi-nology.

KNOWLEDGE ORGANIZATION publishes original articles, re-ports on conferences and similar communications, as well as book re-views, letters to the editor, and an extensive annotated bibliography of recent classification and indexing literature.

KNOWLEDGE ORGANIZATION should therefore be available at every university and research library of every country, at every infor-mation center, at colleges and schools of library and information science, in the hands of everybody interested in the fields mentioned above and thus also at every office for updating information on any topic related to the problems of order in our information-flooded times.

KNOWLEDGE ORGANIZATION was founded in 1973 by an in-ternational group of scholars with a consulting board of editors repre-senting the world’s regions, the special classification fields, and the subject areas involved. From 1974-1980 it was published by K.G. Saur Verlag, München. Back issues of 1978-1992 are available from ERGON-Verlag, too.

As of 1989, KNOWLEDGE ORGANIZATION has become the official organ of the INTERNATIONAL SOCIETY FOR KNOW- LEDGE ORGANIZATION (ISKO) and is included for every ISKO-member, personal or institutional in the membership fee.

Annual subscription 2020: Print + online (8 issues/ann.; unlimited access for your Campus via Nomos eLibrary) € 359,00/ann. Prices do not include postage and packing. Cancellation policy: Termination within 3 months‘ notice to the end of the calendar year

Ergon – ein Verlag in der Nomos Verlagsgesellschaft mbH, Wald-seestraße 3-5, D-76530 Baden-Baden, Tel. +49 (0)7221-21 04-667, Fax +49 (0)7221-21 04-27, Sparkasse Baden-Baden Gaggenau, IBAN: DE05 6625 0030 0005 0022 66, BIC: SOLADES1BAD

Founded under the title International Classification in 1974 by Dr. Ingetraut Dahlberg, the founding president of ISKO. Dr. Dahlberg served as the journal’s editor from 1974 to 1997, and as its publisher (In-deks Verlag of Frankfurt) from 1981 to 1997.

The contents of the journal are indexed and abstracted in Social Sci-ences Citation Index, Web of Science, Information Science Abstracts, INSPEC, Li-brary and Information Science Abstracts (LISA), Library, Information Science & Technology Abstracts (EBSCO), Library Literature and Information Science (Wil-son), PASCAL, Referativnyi Zhurnal Informatika, and Sociological Abstracts.