The effects of topic complexity and familiarity on cognitive and physical moves in a thesaurus-enhanced search environment

http://jis.sagepub.com

Journal of Information Science

DOI: 10.1177/0165551503296008 2003; 29; 517 Journal of Information Science

Ali Asghar Shiri and Crawford Revie thesaurus-enhanced search environment

The effects of topic complexity and familiarity on cognitive and physical moves in a

http://jis.sagepub.com/cgi/content/abstract/29/6/517 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

On behalf of:

Chartered Institute of Library and Information Professionals

can be found at:Journal of Information Science Additional services and information for

http://jis.sagepub.com/cgi/alerts Email Alerts:

http://jis.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://jis.sagepub.com/cgi/content/refs/29/6/517SAGE Journals Online and HighWire Press platforms):

(this article cites 22 articles hosted on the Citations

© 2003 Chartered Institute of Library and Information Professionals. All rights reserved. Not for commercial use or unauthorized distribution. at PENNSYLVANIA STATE UNIV on April 17, 2008 http://jis.sagepub.comDownloaded from

http://www.cilip.org.uk/

http://jis.sagepub.com/cgi/alerts

http://jis.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://jis.sagepub.com/cgi/content/refs/29/6/517


The effects of topic complexity andfamiliarity on cognitive and physicalmoves in a thesaurus-enhancedsearch environment

Ali Asghar Shiri and Crawford Revie

Department of Computer and Information Sciences,University of Strathclyde, Glasgow, UK

Received 16 April 2003Revised 25 July 2003

Abstract.

This paper presents an evaluation of the effects of searchtopic characteristics on cognitive and physical search moveswithin the interface of a thesaurus-enhanced informationretrieval environment. Topic characteristics examined hereare topic complexity, topic familiarity, search type and priortopic search experience. The data gathering techniquesadopted in this investigation included pre- and post-searchquestionnaires, transaction logs and post-session interviews.Thirty academic staff and postgraduate researchers from theFaculty of Veterinary Medicine at the University of Glasgowparticipated in this study. Each participant conducted threesearches based on their research information needs. Theresults show that complex topics are associated withsignificantly more cognitive and physical moves. However,it is perhaps equally important to note that the resultsindicate that variation in the other topic characteristics didnot demonstrate any significant difference in the number ofcognitive or physical moves.

Keywords: online information retrieval; informationseeking behaviour; search strategies; queryformulation; topic complexity; thesaurus use; userinterface; human–computer interaction; end users;experts; novices; comparative studies

1. Introduction

The selection of search terms for query formulationand expansion is a key phase in the informationretrieval process. In particular, thesauri have beenrecognized as a useful source for enhancing searchterm selection [1–3]. Recent developments in end-usersearching and the wider availability of online informa-tion retrieval systems together with advances in user-centred interface design have led to the increased useof thesauri as search aids [4]. The present studyinvestigated the interaction of end-users with athesaurus-enhanced search interface to a large biblio-graphic database on the web. The aim was to evaluateuser interaction in terms of moves made during thesearch process. In particular, the study sought toexplore relations, if any, between the topic character-istics of user queries and the moves they made in athesaurus-enhanced search environment. The presentstudy builds on previous research into: thesauri assearch term sources, information seeking behaviour,and information retrieval interaction.

Correspondence to: A.A. Shiri, Department of Computer andInformation Sciences, University of Strathclyde, 26 Rich-mond Street, Glasgow G1 1XH, UK. E-mail: [email protected]

Journal of Information Science, 29 (6) 2003, pp. 517–526 # CILIP 517 © 2003 Chartered Institute of Library and Information Professionals. All rights reserved. Not for commercial use or unauthorized distribution.

at PENNSYLVANIA STATE UNIV on April 17, 2008 http://jis.sagepub.comDownloaded from


https://www.researchgate.net/publication/235289021_Experiments_on_interfaces_to_support_query_expansion?el=1_x_8&enrichId=rgreq-5f531336-b8a6-46ab-afe2-583194bd9671&enrichSource=Y292ZXJQYWdlOzIzMDYyNjgyMDtBUzoxMDQ2MTQzMzAzNzIxMDdAMTQwMTk1MzQwOTA5MA==

https://www.researchgate.net/publication/220434412_Searchers'_selection_of_search_keys_III_Searching_styles?el=1_x_8&enrichId=rgreq-5f531336-b8a6-46ab-afe2-583194bd9671&enrichSource=Y292ZXJQYWdlOzIzMDYyNjgyMDtBUzoxMDQ2MTQzMzAzNzIxMDdAMTQwMTk1MzQwOTA5MA==

https://www.researchgate.net/publication/230626817_Thesaurus-enhanced_search_interfaces?el=1_x_8&enrichId=rgreq-5f531336-b8a6-46ab-afe2-583194bd9671&enrichSource=Y292ZXJQYWdlOzIzMDYyNjgyMDtBUzoxMDQ2MTQzMzAzNzIxMDdAMTQwMTk1MzQwOTA5MA==

https://www.researchgate.net/publication/2419756_Evaluating_User_Interfaces_to_Information_Retrieval_Systems_A_Case_Study_on_User_Support?el=1_x_8&enrichId=rgreq-5f531336-b8a6-46ab-afe2-583194bd9671&enrichSource=Y292ZXJQYWdlOzIzMDYyNjgyMDtBUzoxMDQ2MTQzMzAzNzIxMDdAMTQwMTk1MzQwOTA5MA==

2. Prior research

Research on searching behaviour, information retrieval(IR) interface evaluation, search term selection andquery expansion has addressed the issue of providingusers with terminological assistance to enhance infor-mation retrieval. Such assistance may be providedthrough the inclusion of thesauri and classificationschemes within IR interfaces.

Researchers have investigated the searching beha-viour of various types of users and have looked inparticular at their search term selection behaviour.Studies have suggested that the selection of terms canbe improved if thesauri are incorporated into thesearch interface [5–8]. A number of researchers haveidentified thesauri as one of the main sources of searchterms in both the query formulation and expansionprocesses [9, 10]. Intelligent interfaces, through whichthe vocabulary of a searcher can be automaticallymatched to the vocabulary of a thesaurus, have alsobeen suggested as being of value [11].

Another line of investigation, providing evidence forthe benefit of incorporating thesauri into interfaces,deals with interactive query expansion and interfaceevaluation. In a series of experiments on designingvarious interfaces to the Okapi search engine, it wasfound that both implicit and explicit use of a thesaurusduring automatic and interactive query expansion wasbeneficial. It was also suggested that while the systemcan find useful thesaurus terms through the queryexpansion process, those terms that were explicitlyselected by users are of particular value [1, 12]. OtherIR interface evaluation studies have found that usersview search term selection as a process which requiresterminological assistance within the interface [2, 13].

In addition to the above empirical studies, a numberof interfaces enhanced with thesauri have been devel-oped over the last two decades [1, 12–20]. Howeververy few of these interfaces have been evaluated interms of the ways in which they support queryformulation and expansion. Nor have users’ percep-tions of issues such as thesaurus interaction andusability in an operational environment been exten-sively addressed.

3. Motivation

The number of commercial thesaurus-enhanced searchinterfaces available to end-users on the web is increas-ing [4]. However, there has been little explanation of

how users interact with these interfaces or of the waysin which end-users’ search term selection for queryformulation or expansion is affected. This study isinterested in specific issues of user-thesaurus interac-tion such as the identification and classification ofmoves made by users during the search process and theeffects on different move types of search topiccharacteristics. A study of search moves is importantbecause it should lead to a better understanding of theinteractive IR process and shed light on cognitive andphysical activities involving users. The search topiccharacteristics examined in this study are discussedbelow.

3.1. Topic complexity

One of the search characteristics examined was topiccomplexity. Borgman et al. [21] have pointed out thatthe concept of the search topic is particularly valuablewhen investigating its effect on search process vari-ables such as search path or behaviour, rather thansimply search outcome variables such as relevancejudgment.

Various approaches have been taken to study thesearch task and its characteristics. These can begenerally categorized as search tasks characteristics,query structure characteristics, and search topic char-acteristics. Hansen [22] has identified a list of sig-nificant search task characteristics and theirimportance in designing interfaces for IR interaction.Some of these task distinctions included: simple vscomplex, active vs passive, and structured vs unstruc-tured. Bystrom and Jarvelin [23] have investigated theeffect of task complexity on information-seekingbehaviour. Research has also been carried out on querystructure and complexity and its effect on informationretrieval. Kristensen and Jarvelin [24] defined querystructure based on the use of operators to express therelations between search keys and have classifiedqueries as having either weak or strong structures.They found that strong query structures lead to betterretrieval performance. In a study of the effect of querycomplexity on web searching results Jansen [25] used asimple vs complex dichotomy based on the number ofsearch terms per query together with the use of Booleanoperators to compare results retrieved from differentWeb search engines. He concluded that increasing thecomplexity of a query had little effect on search results.

Saracevic et al. [26] describe five categories ofquestions based on domain, clarity, specificity, com-plexity and presupposition. They measured the num-ber of search concepts and mapped queries to a five-

Effects of topic complexity and familiarity on cognitive and physical moves in a search environment

518 Journal of Information Science, 29 (6) 2003, pp. 517–526 # CILIP © 2003 Chartered Institute of Library and Information Professionals. All rights reserved. Not for commercial use or unauthorized distribution.



point scale indicating the level of complexity. Theirstudy concluded that questions with higher complex-ity had an increased probability of retrieving morerelevant and precise results. Efthimiadis [5] has alsoinvestigated request characteristics within the contextof an interactive query expansion environment. Thecriteria he used included the subject area of therequest, the nature of the enquiry (accurate or vague),and the type of search being undertaken (broad ornarrow).

In the present study the analysis of topic complexityhas been based on two criteria: the number of searchterms per topic and the number of Boolean operatorsused to transform these terms into a query.

3.2. Topic familiarity, prior topic search experienceand type of search

Research into the effects of topic familiarity on generalsearch behaviour has shown that some informationsearch behaviours such as reading time and topiccomplexity vary with respect to topic familiarity. Kellyand Cool [27] suggest that the more familiar a person iswith a search topic, the less time they spend reading.Hsieh-Yee [6] in studying the effects of subject knowl-edge on the search tactics of novice and experiencedsearchers found that topic familiarity had a significanteffect only for experienced searchers where those witha high level of topic familiarity included moresynonyms and tried more combinations of searchterms.

In the present study topic familiarity was assessedbased on a three-point scale. Users were asked to ratetheir knowledge of the search topic as: very familiar,moderately familiar or unfamiliar. In addition, theusers indicated whether or not they had previouslyperformed a search on each topic. Finally, users wereasked to define the type of search for each topic asbeing either broad or specific. These topic character-istics were all elicited through the pre-search ques-tionnaire.

4. Methodology

The methodology used in a pilot study for the researchreported here was outlined in Shiri et al. [28]. This wasmodified in the light of the results of the pilotexperiment as detailed in the following sections.

4.1. System

The CAB Abstracts database, the largest bibliographicdatabase in the agricultural sciences, and provided byOvid on the web, was utilized as the experimentalplatform. The Ovid search interface was chosen as itserves the purpose and requirements of the presentinvestigation. Ovid CAB has a thesaurus-enhancedsearch and browse feature in its advanced search modewhich allows users to map their search terms to theCAB thesaurus. The interface, as shown in Fig. 1,makes use of the thesaurus to provide an enhancedsearch environment. Depending on the users’ initialquery, the interface shows one of the following threematch types to users who may then browse and selectterms for querying the CAB Abstracts database:. exact match – the user’s term matches exactly one

of the thesaurus terms;. permuted index match – the user’s term matches

partially one of the thesaurus descriptors which arearranged in a permuted index;

. statistical match – through statistical analysis thedescriptors frequently co-occurring with the user’sterms are shown.

4.2. Participants

The purpose of this study was to explore the termselection and thesaurus interaction behaviour of realusers with genuine information needs. To this end, itwas decided to involve researchers from a particularknowledge domain. As Table 1 shows, the participantsin the experiment were 30 faculty and postgraduateresearchers, who were selected from the Faculty ofVeterinary Medicine at the University of Glasgow. Thisgroup of participants was targeted due to the fact thattheir research interests are well covered within theCAB Abstracts database. Fifteen users were facultymembers and the rest were postgraduates and doctoralstudents. Most students were ‘clinical scholars’ whichmeans they were involved in veterinary practice as

Table 1.Participants grouped by gender and status

Status Gender Total

Male Female

Faculty member 7 8 15

Postgraduate/PhD students 6 9 15

Total 13 17 30

A.A. SHIRI AND C. REVIE




well as research. The participants were self-selectedthrough e-mail contact.

4.3. Search tasks

There are two main types of search tasks reported inthe literature which may be adopted depending on theaims and objectives of a study: assigned search tasksand topics usually defined by the researcher [2, 6, 29,30], or real search topics elicited from users in the lightof genuine information needs [5, 12, 31]. Borlund andIngwersen [32] have also introduced the concept of thesimulated information need situation.

In the present study, search topics provided by theparticipants were used to create an environmentwithin which the interaction with the thesaurus andinterface was as natural as possible. This decision wasmade on the assumption that the evaluation of searchterm selection, query expansion and users’ interactionwith the thesaurus can be more effectively carried outif users with genuine queries are involved in the study.

The subjects were asked to provide, in a pre-searchquestionnaire, three search topics on which theywould like to conduct searches. They were also askedwhether they had previously carried out searches oneach topic.

4.4. Data gathering techniques

The complex nature of user interaction with IR systemscalls for the application of a combination of variousdata gathering techniques. This study employed bothquantitative and qualitative techniques to collectinteraction data. A pre-search questionnaire wasutilized to capture some user background in additionto the three search topics. Lotus ScreenCam softwarewas used as a monitoring tool to capture the completeuser interaction with the interface. A post-searchquestionnaire elicited a response from users on thesearch terms and results provided for each topic. Apost-session interview was also carried out to gatherinformation on users’ general impressions about the

Fig. 1. Ovid thesaurus interface.





thesaurus, the interface and their interaction experi-ence.

4.5. Experimental procedures

Users came to attend the experiment one at a time.They were initially given a brief description of theresearch objectives and the stages of the experiment. Ashort written tutorial was provided to ensure consis-tency in the introduction to the system. The tutorialprovided users with a short description of thedatabase, the search facilities and in particular thethesaurus mapping facility which was available in theadvanced search mode. All users performed a practicesearch to become familiar with the search interface andthe thesaurus mapping facility. The practice searchwas designed in such a way as to provide users with anopportunity to gain experience of a typical thesaurus-based search session and to demonstrate the manner inwhich thesaurus terms and structure are presentedwithin the OVID interface. Users were then encouragedto use the thesaurus while conducting three searchesbased on the topics they had specified in the pre-searchquestionnaires. After each search users filled in a post-search questionnaire. An interview was also conductedat the end of the session.

5. Description of key variables

Two types of variables were defined in this study. Theindependent variables included topic complexity,topic familiarity, search type and prior topic searchexperience. The dependent variables were cognitiveand physical moves. In the following a description ofthese variables and the ways in which they weredefined is provided.

5.1. Topic complexity

In order to measure the complexity of the topicssearched for by users, the number of search termsand number of operators used in each topic wereassessed. For the purpose of comparative analysis thetopics were classified as either simple or complex.Topics having more than three search terms or morethan two operators were classified as complex andthose with three search terms or less and two operatorsor less were categorized as simple topics. Thisclassification led to around 60% of the searches beingdefined as simple.

5.2. Topic familiarity and prior topic searchexperience

The total number of topics within each familiarity leveland prior experience grouping is indicated in Table 2.

5.3. Search type

Users were asked to determine whether they intendedto conduct a broad or specific search. The datagathered through this question was used to analysewhether the type of search had any effect on the movesusers made while interacting with the interface. Justover one third of the topics were stated by the users tobe specific.

5.4. Search moves

Various approaches have been taken to define searchmoves in studying online search behaviour. Forinstance Bates [33] introduced a number of searchmoves or tactics some of which specifically relate tosearch formulation and search terms. Fidel [34]identified two types of online search move whenstudying professional searchers, namely operationaland conceptual moves. Marchionini et al. [35] in

Table 2.Topics grouped by familiarity level and prior search experience

Topic familiarity Prior topic search experience Total

Not searched previously Searched previously

Unfamiliar 8 7 15Moderately familiar 20 32 52Very familiar 5 18 23Total 33 57 90





studying user interaction with IR interfaces defined amove as, ‘a conceptual act manifested as one or morekeystrokes’, for instance entering a query string orpressing the page-down key.

Since the theoretical framework underlying thepresent study focused on cognitive aspects of userinteraction with IR systems two general categories ofmoves were defined for the purpose of analysis:. cognitive moves – in which users perform some

kind of conceptual analysis of terms or documents;. physical moves – those associated with the use of

system features.Studying these types of move is important as it

provides insight into various layers of user interactionand shows the extent to which each move contributesto the whole search process.

Cognitive moves. Seven cognitive move types wereidentified in the search process. These are shown inTable 3 together with the total number of moves of eachtype and the mean number per search.

Users can browse thesaurus terms in two ways.Upon inputting a term, a list of exact, partial, orstatistically matched terms is displayed. This allowsusers either to browse the list or click on a particularthesaurus term to see the details of the term in ahierarchical display (a less commonly used approachas can be seen from Table 3). Combine search termsindicates the number of times users made a decision asto which search terms would be combined usingBoolean operators. Query reformulation encompassesany effort to modify, enhance or expand the query.

Physical moves. In total 10 physical move types wereidentified. The total and mean number of moves persearch are shown in Table 4. The Back and forwardcategory refers to the browser’s facility for backtrackingor going forward, while Previous and next page arefacilities for flipping through result pages. TheContinue move type is invoked when users want toproceed with the search process from the mappingpage, the thesaurus page and also the combine search

Table 3.Total and mean number of cognitive moves per search

Cognitive move types Total number of moves Mean number of moves per search

Term input 252 2.8Browsing terms in mapping state 210 2.4Browsing terms in thesaurus hierarchy state 55 0.6Selection of terms 265 2.9Combine search terms 192 2.1Browsing retrieved titles 111 1.2Query reformulation 104 1.1

Table 4.Total and mean number of physical moves per search

Physical move types Total number of moves Mean number of moves per search

Perform search 254 2.8Scroll up and down 671 7.5Back and forward 39 0.5Continue 192 2.1Combine 331 3.7Citation display and e-mail 284 3.2Main search page 10 0.1Previous and next page 71 0.8Search history 15 0.2Expand and contract 38 0.4





page. The Combine category denotes the number oftimes the user clicked the icon which activated one ofthe Boolean operators.

6. Results

6.1. Relationship of topic characteristics to physicaland cognitive moves

The topic-related variables were analysed to investi-gate possible relationships between topic characteris-tics and the user’s search moves. The results aresummarized in the succeeding sections.

Topic complexity and cognitive moves. To evaluatethe effect of topic complexity on cognitive moves madeduring the search process a t test was conducted. Sinceoutliers have an undue influence on the generaldistribution of data, four searches identified asoutliers in terms of their number of cognitive moveswere excluded to provide a normally distributed dataset. As shown in Table 5, complex topic searchesinvolved around six more cognitive moves per searchthan those involving simple topics. The t-test showedthis difference to be highly significant (p<0.001), andthe confidence interval (CI) indicates that there is a95% probability that the difference lies in the range of4.9–7.9 moves. This finding is in line with thatreported by Marchionini [36] who found that thenumber of moves was dependent on the task, andthat more complex tasks lead to more moves beingmade during a search. Fowkes and Beaulieu [37] alsofound that complex topics required a higher level ofengagement and effort in areas such as interactivequery expansion and the reading of retrieved results.

Topic complexity and physical moves. The differencebetween the mean number of physical moves withrespect to simple and complex topics was alsocompared using a t-test. Two outliers were excludedto avoid bias when considering physical moves. Table 6shows that complex topic searches were associated

with an average of around seven more physical movesper search than for simple topics. Once again thedifference was highly significant (p<0.001) and the95% confidence interval indicates the typical range ofvalues this difference would fall within. The fact thatthe level of complexity affects the number of physicalmoves is consistent with the findings of Marchionini[35] who reported that complex tasks cause users tomake more use of system features.

Topic familiarity, topic search experience and searchtype. The number of search topics excluding outliersand mean number of cognitive and physical moves foreach level of topic familiarity are shown in Table 7.

A one-way ANOVA test indicated that there was nosignificant difference between topic familiarity levelsin either the number of physical (p¼ 0.07) or cognitive(p¼ 0.3) moves. However, it can be seen that topicsidentified as being moderately or very familiar wereassociated with around two more cognitive and fivemore physical moves than those topics identified asunfamiliar. These findings can be compared to thosereported by Hsieh-Yee [6] who found that subjectknowledge affected the searching behaviour of experi-enced searchers.

The effects on cognitive and physical moves of priortopic search experience and search type were exam-ined. The results are shown in Tables 8 and 9.

A t-test was carried out to confirm that there was norelationship between prior topic search experience andcognitive or physical moves. The same test alsoindicated no significant difference between the means

Table 5.Mean number of cognitive moves by topic type

Complexity level n Mean cognitive moves CI

Simple topics 51 9.9(4.9, 7.9)

Complex topics 35 16.2

Table 6.Mean number of physical moves by topic type

Complexity level n Mean physical moves CI

Simple topics 51 17.3(4.4, 10.5)

Complex topics 37 24.7

Table 7.Mean number of moves by topic familiarity

Level of topic Cognitive moves Physical movesfamiliarity Mean n Mean n

Unfamiliar 10.8 14 16.2 15Moderate 13.0 49 21.7 50Very familiar 12.3 23 20.5 23





of cognitive or physical moves for broad and specificsearch types. These findings suggest that neither searchtype nor previous topic search experience affected thenumber of moves users made during the searchprocess.

6.2. Relationships between independent variables

To test whether there were any relationships betweenthe independent variables; namely topic complexity,topic familiarity, search type and prior topic searchexperience, chi-square tests were carried out. Norelationship was found between any of these variables.Of interest was the finding that there was no relation-ship between topic familiarity and type of search. Thiscan be seen as contradicting the generally heldassumption that the more familiar a user is with atopic, the more specific their search will be. Thefinding that topic familiarity was unrelated to topiccomplexity can also be contrasted to the case of expertsearchers reported in Hsieh-Yee [6].

7. Conclusion

This study investigated the effects of topic complexity,topic familiarity, topic search experience and searchtype on cognitive and physical moves in a thesaurus-enhanced online operational IR system. The findingsindicate that an increased number of cognitive andphysical search moves were associated with morecomplex topics. It was also observed that users

searching moderately familiar and very familiar topicsused more moves than users searching for unfamiliartopics, though this difference was not statisticallysignificant. Type of search and prior topic searchexperience were found not to affect the number ofcognitive or physical moves.

Analysis which adopts this approach to the classi-fication of search moves can be of value in the design ofIR interfaces. In addition to the identification of majorcategories of move and their relative frequencies, theapproach facilitates an evaluation of the contributionmade by each type of move to the overall searchprocess and their importance within the broader set ofinteractive IR evaluation measures. Results from thisstudy indicate that complex topic searches requiremore query reformulation and involve users in viewingmore search results. The integration of search term andsearch result spaces could be viewed as one mechan-ism for improving the interface to better supportcognitive moves when performing complex topicsearches. Another implication for interface designwould be that facilities for query modification andexpansion should be made more readily accessible tousers carrying out searches associated with complextopics.

When compared with search outcome measures suchas relevance judgement or user satisfaction with searchresults, cognitive and physical moves can be viewed asmeasures contributing to the evaluation of IR interac-tion and search efficiency. Furthermore, the introduc-tion of cognitive and physical moves provides amethod for assessing the extent to which users takeadvantage of browse, search and navigation featureswithin an IR system, and can guide the interface designprocess such that cognitive and physical loads arereduced during a search process.

References

[1] M. Beaulieu, Experiments of interfaces to support queryexpansion, Journal of Documentation 53(1) (1997) 8–19.

[2] G. Brajnik, S. Mizzaro and C. Tasso, Evaluating userinterfaces to information retrieval systems: a case studyon user support. In: Proceedings of the 19th AnnualInternational ACM/SIGIR Conference on Research andDevelopment in Information Retrieval, (Hartung-GorreVerlag Konstanz, Germany, 1996), pp. 128–136.

[3] R. Fidel, Searchers’ selection of search keys: III.Searching styles, Journal of the American Society forInformation Science 42(7) (1991) 515–527.

Table 8.Mean number of moves by previous topic search experience

Topic search Cognitive moves Physical movesexperience Mean n Mean n

Experienced 12.7 55 20.8 55No experience 12.1 31 19.8 33

Table 9.Mean number of moves by search type

Search type Cognitive moves Physical movesMean n Mean n

Broad 11.9 54 21.3 55Specific 13.4 32 19.0 33





[4] A. Shiri, C. Revie and G. Chowdhury, Thesaurus-enhanced search interfaces. Journal of InformationScience 28(2) (2002) 111–122.

[5] E.N. Efthimiadis, Interactive query expansion: a user-based evaluation in a relevance feedback environment,Journal of the American Society for Information Science51(12) (2000) 1–25.

[6] I. Hsieh-Yee, Effects of search experience and subjectknowledge on the search tactics of novice and experi-enced searchers, Journal of the American Society forInformation Science 44(3) (1993) 161–174.

[7] A.G. Sutcliffe, M. Ennis, and S.J. Watkinson, Empiricalstudies of end-user information searching, Journal of theAmerican Society for Information Science 51(13) (2000)1211–1231.

[8] P. Vakkari, Cognition and changes of search terms andtactics during task performance: a longitudinal casestudy. In: RIAO 2000 Content-based Multimedia Infor-mation Access, Paris (12–14 April 2000), pp. 894–907.

[9] A. Spink, Term relevance feedback and mediateddatabase searching: implications for information retrie-val practice and systems design, Information Processingand Management 31(2) (1994) 161–171.

[10] A. Spink and T. Saracevic, Interaction in informationretrieval: selection and effectiveness of search terms,Journal of the American Society for Information Science48(8) (1997) 741–761.

[11] R. Fidel, Searchers’ selection of search keys: III.Searching styles, Journal of the American Society forInformation Science 42(7) (1991) 515–527.

[12] S. Jones, M. Gatford, S. Robertson, M. Hancock-Beaulieu, J. Secker and S. Walker, Interactive thesaurusnavigation: intelligence rules Ok? Journal of the Amer-ican Society for Information Science 46(1) (1995) 52–59.

[13] C.F. McMath, , R.S. Tamaru and R. Rada, A graphicalthesaurus-based information retrieval system, Interna-tional Journal of Man–Machine Studies 31 (1989) 121–147.

[14] M. Agosti, G. Gradenigo and P.G. Marchetti, A hypertextenvironment for interacting with large textual databases,Information Processing and Management 28(3) (1992)371–387.

[15] N.J. Belkin, P.G. Marchetti and C. Cool, BRAQUE:design of an interface to support user interaction ininformation retrieval, Information Processing and Man-agement 29(3) (1993) 325–344.

[16] E. Johnson and P. Cochrane, A hypertextual interface fora searcher’s thesaurus. In: Digital Libraries ‘95: Proceed-ings of the Second Annual Conference on the Theoryand Practice of Digital Libraries, Austin, TX (11–13 June1995).

[17] R. Pollard, A hypertext-based thesaurus as subjectbrowsing aid for bibliographic database, InformationProcessing and Management 29(3) (1993) 345–357.

[18] A.S. Pollitt, G.P. Ellis and M.P. Smith, HIBROWSE forbibliographic databases, Journal of Information Science20(6) (1994) 413–426.

[19] X. Lin, Visual MeSH. Proceedings of 22nd AnnualInternational ACM SIGIR Conference on Research anddevelopment in Information Retrieval, Berkeley, CA,15–19 August 1999 (ACM, New York, 1999), pp. 317–318.

[20] M. Hearst and C. Karadi, Cat-a-Cone: an interactiveinterface for specifying searches and viewing retrievalresults using a large category hierarchy. In: SIGIR ‘97:Proceedings of the 20th Annual International ACM/SIGIR Conference on Research and Development inInformation Retrieval, Philadelphia, PA, USA, 27–31July 1997 (ACM, New York, 1997), pp. 246–255.

[21] C. L. Borgman, S.G. Hirsh and J. Hiller, Rethinkingonline monitoring methods for information retrievalsystems: from search product to search process Journalof the American Society for Information Science 47(7)(1996) 568–583.

[22] P. Hansen, User interface design for IR interaction: atask-oriented approach. In: T. Saracevic, P. Ingwersenand P. Vakkari (eds), CoLIS 3. Proceedings of the ThirdInternational Conference on the Conceptions of theLibrary and Information Science, Dubrovnik, Croatia(23–26 May 1999), pp. 191–205.

[23] K. Bystrom and K. Jarvelin, Task complexity affectsinformation seeking and use, Information Processingand Management 31(2) (1995) 191–213.

[24] J. Kristensen, J. and K. Jarvelin, The impact of querystructure and query expansion on retrieval performance.In: W.B. Croft, A. Moffat, C.J. van Rijsbergen, R.Wilkinson and J. Zobel (eds), SIGIR’98: Proceedings ofthe Association for Computing Machinery SpecialInterest Group on Information Retrieval (ACM/SIGIR)21st Annual International ACM SIGIR Conference onResearch and Development in Information Retrieval,Melbourne (ACM Press, New York, 1998), pp. 130–137.

[25] B. Jansen, The effect of query complexity on Websearching results, Information Research, 6(1) Availableat: http://InformationR.net/ir/6–1/paper87.html (2000).

[26] T. Saracevic, P. Kantor, A. Chamis and D. Trivison, Astudy of information seeking and retrieving. I. back-ground and methodology, Journal of the AmericanSociety for Information Science 39(3) (1988)161–176.

[27] D. Kelly and C. Cool, The effects of topic familiarity oninformation search behaviour. In: Proceedings of JointConference on Digital Libraries (JCDL) (13–17 July 2002),pp. 74–77.

[28] A. Shiri, C. Revie and G. Chowdhury, Assessing theimpact of user interaction with thesaural knowledgestructures: a quantitative analysis framework. In: Chal-lenges in Knowledge Representation and Organizationfor the 21st Century: Integration of Knowledge AcrossBoundaries: Proceedings of the 7th International Society





for Knowledge Organization (ISKO) Conference, Gran-ada, Spain (10–13 July 2002), pp. 493–499.

[29] C. Meadow, J. Wang and W. Yuan, A study of userperformance and attitudes with information retrievalinterfaces, Journal of the American Society for Informa-tion Science 46(7) (1995) 490–505.

[30] A.G. Sutcliffe, M. Ennis and S.J. Watkinson, Empiricalstudies of end-user information searching, Journal of theAmerican Society for Information Science 51(13) (2000)1211–1231.

[31] J. Greenberg, Optimal query expansion (QE) processingmethods with semantically encoded structures thesauriterminology, Journal of the American Society forInformation Science and Technology 52(6) (2001) 487–498.

[32] P. Borlund and P. Ingwersen, The development of amethod for the evaluation of interactive informationretrieval systems, Journal of Documentation 53(3) (1997)225–250.

[33] M. Bates, Information search tactics Journal of theAmerican Society for Information Science 30(4) (1979)205–214.

[34] R. Fidel, Moves in online searching, Online Review 9(1)(1985) 61–74.

[35] G. Marchionini, C. Meadow, S. Dwiggins, X. Lin, J.Wang and W. Yuan, A study of user interaction withinformation retrieval interfaces: progress report, TheCanadian Journal of Information Science 16(4) (1991)42–59.

[36] G. Marchionini, Information-seeking strategies ofnovices using a full-text electronic encyclopedia, Jour-nal of the American Society for Information Science40(1) (1989) 54–66.

[37] H. Fowkes and M. Beaulieu, Interactive searchingbehaviour: Okapi experiment for TREC-8. In: TheBCS/ IRSG 22nd Annual Colloquium on InformationRetrieval Research, Cambridge (5–7 April 2000),pp. 47–56.





The effects of topic complexity and familiarity on cognitive and physical moves in a thesaurus-enhanced search environment

Documents