Semantic Interoperability and Retrieval Paradigms
Paradigms and conceptual systems in KOFebruary 23, 2010 – February 26, 2010
Prof. Winfried GödertFelix Boteram Jessica Hubrich
2
Retrieval Focussed search Expansion Exploration
3
Classifications Thesauri Semantic networks Ontologies
Knowledgeorganization
systems
Retrieval
Interoperability
4
Knowledgeorganization
systems
Retrieval
Technical Structural Semantic
Semantic interoperability
5
Semantic interoperability
6
Retrieval
7
Aims
Strategies for retrieval and exploration in distributed systems Integration of structurally and typologically different systems Overall system consisting of an ontological spine and attached
satellites
Features
Simple hierarchical core system Detailed bilateral mappings Expressive multidimensional definition of intersystem and
interconcept relations Shared inventory of standard relations
Overall model for semantic Interoperability
8
SWSWD SWLCSH
SWRameau
SWS
SWS
SWS
Syn.SWL
SWL
Syn.
SWL
SWR
SWR
SWRSyn.
SWS
SWxy
SWxy
SWxy
SWxySyn.
SWxy
SWR
SWL
Localized semantic networkswith typed relations as link
to the ontological spine
Core System*)
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
possible extensionswith satellites
9
Retrieval paradigms
Simple pattern matching
Thematic exploration
Conceptual query
Conceptual exploration
„TREE“
10
Levels of interoperability
Basic interoperability
Conceptual interoperability
Enhanced conceptual interoperability
„TREE“
11
Features
Easy implementation No indexing languages required Limited access and limited search functionalities No modification / expansion possible
Methods
Mapping on word level Unspecified one-to-one equivalences represented in a
simple cross-concordance.
Basic interoperability
Simple pattern matching „TREE“
12
Basic interoperability
Simple pattern matching
13
Features
Focus on the conceptual level Multiple access points via synonyms Structural transparency Simple expansion strategies
Methods
Mapping concepts independently from their actual representation in the various indexing languages.
Structural and typological differences should be taken into account.
Conceptual interoperability
Conceptual query
14
Conceptual interoperability
Conceptual query
15
Features
Modification / expansion of queries Conceptual exploration of a knowledge field Query-clarification as a useful pre-search mechanism
Requirements
Various types of expressive semantic relations have to be specified in great detail.
Well-structured relational inventories Control- and selection-mechanisms to handle the complexity of
the relational structure
Enhanced conceptual interoperability
Conceptual exploration
16
Methods
Differentiated conceptual mappings Interpretation of the semantically enhanced relational
structure Relations between concepts have to be specified
Enhanced conceptual interoperability
Conceptual exploration
17
Enhanced conceptual interoperability
Conceptual exploration
18
19
20
21
SWSWD SWLCSH
SWRameau
SWS
SWS
SWS
Syn.SWL
SWL
Syn.
SWL
SWR
SWR
SWRSyn.
SWS
SWxy
SWxy
SWxy
SWxySyn.
SWxy
SWR
SWL
Localized semantic networkswith typed relations as link
to the ontological spine
Core System*)
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
Syn.
possible extensionswith satellites
22
Thank you for your attention
Prof. Winfried Gö[email protected]
Felix BoteramJessica Hubrich
Institute of Information Management (IIM)Cologne University of Applied Sciences
Thematic exploration / thematic interoperability ?
Thematic exploration requires navigation and interpretation on the conceptual level via an enhanced conceptual interoperability and an exploratory component on the syntactical level.
Thematic entities / Syntagmata are always combinations of semantic entities and syntactic rules.
Need for a clear-cut distinction between the enhanced semantic level and the syntactic level.
Thematic exploration / thematic interoperability ?
Syntagmata can only be accessed by a careful analysis of their semantic components and syntactic structures.
Themes / Syntagmata can never be part of the semantic level, therefore the idea of a thematic interoperability as an advanced level of semantic interoperability is problematic.
Semantic interoperability between individual KOS can function as an intermediary when accessing / matching syntagmata.
25
From retrieval to knowledge exploration: example 1
Data basis: ca. 330.000 documents of the German National Library (DNB) which are heterogeneously indexed (DDC, SWD)Search interest: GemüseanbauSearch result Subject
headingTitle Full text
0 6 13Semantic environment: Gemüseanbau -> GemüsebauSearch result Subject
headingTitle Full text
11 14 31Integration of all narrower SWD terms 20 hits
From retrieval to knowledge exploration: example 1
Transition to DDCGemüsebau (Register) -> 635 Gartenpflanzen (Gartenbau) ; Gemüse
Search for 635: 198 hits (as notation)351 hits (as part of a built number)
600 Technik, Medizin, angewandte Wissenschaften 630 Landwirtschaft 633-635 Einzelne Kulturpflanzen 635 Gartenpflanzen (Gartenbau) ; Gemüse 635.0284 Geräte, Ausstattung, Materialien 635.0285 Gartenbau—Computereinsatz 635.029 Gartenbau - Unternehmen - Verzeichnis 635.04 *Anbau, Ernte, verwandte Themen 635.05 Gartenbau—fortlaufende Sammelwerke 635.1-635.8 Gartengemüse 635.9 Blumen und Zierpflanzen
Context:
From retrieval to knowledge exploration: example 1
Connected to the class are:12 index terms 4 SWD headings of high relevance27 SWD headings of average relevance8 SWD headings of low relevance
GemüsebaubetriebBlattfrüchte Exotisches GemüseFeldgemüse Feldgemüsebau FrühgemüseGemüse Gemüsebau GemüsegartenGemüseproduktion
high
average
low
Selection:
From retrieval to knowledge exploration: example 1
Option 1:Structure search results according to the mapped SWD
headings
From retrieval to knowledge exploration: example 1
Option 2:Modified queries by selection of mapped SWD headings, for example:Search for: Exotisches Gemüse: 1 hit
Feldgemüse: 1 hitGemüse: 36 hitsGemüsebau: 11 hitsGemüsegarten: 23 hitsGemüseproduktion: 1 hit
31
From retrieval to knowledge exploration: example 2
Data basis: ca. 330.000 documents of the German National Library (DNB) which are heterogeneously indexed (DDC, SWD)Search interest: JagenSearch results: Subject
headingTitle Full text
0 15 202Semantic environment: Jagen -> JagdSearch results: Subject
headingTitle Full text
127 102 524Integration of all narrower SWD terms : 160 hits
From retrieval to knowledge exploration: example 2
Transition to DDCJagd (Register) -> 639.1 Jagd [Alternative: 799.2]
Search for 639.1: 27 hits (as notation)137 hits (as part of built number)
Search for 639.1*: 162 hits
630 Landwirtschaft 639 Jagd, Fischfang, Erhaltung, verwandte Bereiche 639.1 Jagd 639.11 Jagd auf Säugetiere 639.12 Jagd auf Vögel 639.13 Jagd auf Amphibien
639.14 Jagd auf Reptilien
Context:
From retrieval to knowledge exploration: example 2
Connected to class 639.1 are:6 index terms 3 SWD headings of high relevance12 SWD headings of average relevance8 SWD headings of low relevance
JagdAbschuss AbschussplanAnsitzBeutefangBrackierenHegeNachsucheRufjagdWildbestand
high
average
low
Selection:
From retrieval to knowledge exploration: example 2
Transition to DDCJagd (Register) -> 799.2 Jagd [Alternative: 639.1]
Search for 799.2: 4 hits (as notation)12 hits (as part of a built number)
Search for 799.2*: 37 hits
790 Sport, Spiele, Unterhaltung 799 Fischfang, Jagd, Schießen 799.2 Jagd 799.202 Verschiedenes 799.21-799.23 Methoden 799.24-799.27 Jagd auf einzelne Arten von Tieren
799.29 Historische, geographische, personenbezogene Behandlung
Context:
From retrieval to knowledge exploration: example 2
Connected to the class 799.2 are:4 index terms 3 SWD headings of high relevance18 SWD headings of average relevance12 SWD headings of low relevance
JagdAbschuss AbschussplanBalgBeutefangGesellschaftsjagd JagdunfallTrophäe (Jagd) WaldjagdWildbestand
high
average
low
Selection: