KOSO: A Reference-Ontology for Reuse of Existing Knowledge Organization Systems International Workshop on Knowledge Reuse and Reengineering over the Semantic Web (KRRSW 2008) ESWC 2008 Tenerife, Spain, June 2008 Katrin Weller Institute for Language and Information, Dept. of Information Science, Heinrich-Heine-University Düsseldorf KRRSW at ESWC 2008, Tenerife
Research Questions and Objective Where do I find existing knowledge models that can be reused? How can ontologies be classified and described? What are essential properties of ontologies? Meta-level of of ontology engineering How are different knowledge organization systems distinguished? How do other knowledge representations relate to ontologies? How can I retrieve specific ontologies? Has a certain ontology been reused? Has this ontology connections to others? 2
Motivation and Current Situation Reuse of knowledge models for ontology engineering - To save time and manpower. - To transfer existing knowledge. Plenty of different domain models are already available for different purposes. Problems: Different underlying principles, different degrees of formalization, varying quality. No precise overview but single approaches to make existing ontologies accessible. 3
Example I Web Search Engine Approach: Swoogle 4
Example II Web Repository Approach: SchemaWeb 5
Examples III Domain specific collection: OBO Format specific collection: DAML Library 6
Metadata for Ontologies Metadata Approaches: Enabling consistent metadata descriptions for ontologies. Most prominent: OMV Ontology Metadata Vocabulary Source: OMV_v2.4 http://omv.ontoware.org/ 7
KOSO Knowledge Organization Systems Ontology New Approach: KOSO Knowledge Organization Systems Ontology Focus on: Different types of knowledge organization systems (KOS) Their characteristics, particulary the use of knowledge relations Their (possible) interactions Question: What is an ontology? In this approach: it is the most complex available KOS. Work in progress: suggestions, questions, comments? 8
Initial Structure Starting with classical and modern KOS. Future aim: broadening to other knowledge resources (KnowledgeResouce as upper class for KnowledgeOrganizationSystem, LinguisticResource, EncyclopedicResource.) 9
Other Knowledge Resources Aim: Consideration of different types of knowledge resources, not only existing ontologies. Ontology Nomenclature Folksonomy Classification Thesaurus Current focus Dictionary Linguistic thesaurus Lexicon Wiki Glossary Web Directory 10
Initial Structure Starting with KOS developed to index documents. Future aim: analyzing other purposes and application scenarios. 11
Initial Structure Focus on semantic relations as key elements for distinguishing and upgrading KOS. Future aim: more detailed description of other KOS elements. 12
Semantic Relations 13
Semantic Relations in KOSO 14
KOS Types by Semantic Relations Semantic upgrades 15
Semantic Upgrades: Example Tag Gardening Identifying and relating hyponyms and hypernyms (is_a). Identifying and relating meronyms and holonyms (part_of). Identifying and relating synonyms and quasi-synonyms (same_as). 16
Defining KOS Types - Overview Ontology Explicitly specified semantic relations. Formal representation language (for automatic reasoning). Distinguishing concepts and individuals. Example: Cyc Thesaurus Focus on elaborated vocabulary control: meronymy, hyponymy, equivalences and unspecific associative relations. Example: Medical Subject Headings (MeSH) Classification Mainly hierarchical structure, equivalence relations. Subtypes, e.g. decimal classification, faceted classification. Uses notations. Example: International Patent Classification (IPC) Nomenclature Controlled keyword indexing with focus on equivalence relations (synonyms), additional associations possible. Example: CAS Registry File Folksonomy No concept interrelations. Developed by community, bound to platform. Subtypes: broad and narrow folksonomy. Examples: Flickr Folksonomy, Del.icio.us Folksonomy. 17
KnowledgeOrganizationSystem in KOSO 18
KnowledgeOrganizationSystem in KOSO 19
Properties to Describe KOS Properties to interrelate different elements of the ontology. Additional datatype properties regarding size, availability, specific features etc. Examples: has_number_of_concepts is_polyhierarchic 20
Example International Patent Classification (IPC) 21
Example International Patent Classification (IPC) is_a Classification (is_a KnowledgeOrganisationSystem) has_version: IPC2006_english, IPC2006_french, available_in_language: English, French uses_notations = true is_developed_by: WorldIntellectualPropertyOrganization has_number_of_concepts: 50,000-100,000 uses_relation: Hierarchy has_domain: IntellectualProperty is_used_to_index: InternationalPatents 22
Types of Interactions and Interrelations I Modifications & Versioning: Interlinking different release versions, modified versions, language versions of the same KOS. Modularization: Example: Gene Ontology consists of three modules Biological Process, Cellular Component and Molecular Function. Discussion: Types of modules? Facets, top level concepts? Reuse: Reuse of a given KOS to develop a new one. May be specified to complete and partly reuse. Example: SmartSUMO ontology reuses DOLCE and SUMO. 23
Types of Interactions and Interrelations II Cross-References Established cross-references between concepts of two different KOS. Mapping of two KOS. Example: Standard Thesaurus Wirtschaft has_concordances_to NACE. Special case: informal quotations/references to single concepts within another KOS. Multirepresentation: Independent KOS are used within the same platform. They may represent different points of view. Example: Documents in one platform are professionally indexed with a thesaurus and additionally tagged with a community s folksonomy. 24
Interactions of KOS: Properties in KOSO Modification & Versioning Modularization has_version (1:n) is_prior_version_of (1:1) is_later_version_of (1:1) is_language_version_of consists_of_module and inverse is_component_of Resources and Reuse is_resource_for and inverse reuses (semantic_upgrade_of) Cross-references has_concordances_to (quotes) Multi-representation used_in_combination_with (symmetric) 25
Example KOSO has_version: KOSO-2008-06-01_english is_later_version_of: KOSO-2008-03-15_english consists_of_module: KnowledgeResource, Domain, Platform, reuses: LanguageModule (is_component_of Thesaurofacet) quotes: OMV (is_a Ontology), ASIST_Thesaurus (is_a Thesaurus), Glossary_of_Terminology_Wellisch_2003 (is_a Glossary) is_resource_for: xyz 26
Outlook: Extending KOSO Inclusion of more types of knowledge resources more tasks and application types engineering methodologies and tools standards and norms (e.g. ISO 2788-1986 for monolingual thesauri) licence types Check for other reusable resources to integrate into KOSO. Evaluation and Application. Alternatively: integrating certain KOSO elements to existing approaches? 27
Conclusion & Discussion KOSO provides definitions for different KOS. Discussion: Do they represent a shared view? Should they be remodeled? KOSO regards knowledge relations as important characteristic of KOS. Discussion: Appropriateness? Other ideas? KOSO represents different types of KOS interactions. Discussion: Are there other types? Should the types be refined? 28
Thank You! Greetings from Düsseldorf Questions, suggestions, comments? Interested in a copy of the latest KOSO version? Contact me: weller@uni-duesseldorf.de Katrin Weller Institute for Language and Information Dept. Information Science Heinrich-Heine-University Düsseldorf Universitätsstraße 1, Building 23.21.04 D-40225 Düsseldorf 29
References (see proceedings for more) Aitchison, J., Gomersall, A., Ireland, R.: Thesaurofacet: A Thesaurus and Faceted Classification for Engineering and Related Subjects. Rugby, England: Jolly and Barber (1969) Arpirez, J. C., Gómez-Pérez, A., Lozano-Tello, A., Pinto, H. S.: Reference Ontology and (ONTO)2 Agent: The Ontology Yellow Pages. Knowledge and Information Systems, 2(4), 387--412 (2000) Hartmann, J., Suarez-Figueroa, M. Carmen, Sure, Y., Haase, P., Palma, R.: OMV. Ontology Metadata Vocabulary. In C. Welty (Ed.), ISWC 2005. Ontology Patterns for the Semantic Web (2005) Hartmann, J.: ONTHOLOGY. An Ontology Metadata Repository. In Demo and Poster Proceedings of ESWC 2006 (2006) Peters, I.; Weller, K.: Paradigmatic and Syntagmatic Relations in Knowledge Organization Systems. In: Information Wissenschaft und Praxis 59(2), pp. 100--107 (2008) Redmond-Neal, A., Hlava, M. M. K. (Eds.): ASIS&T Thesaurus of Information Science, Technology, and Librarianship. 3rd ed. Medford, NJ, USA: Information Today (2005) Suarez-Figueroa, M. C., García-Castro, R., Gómez-Pérez, A., Palma, R., Nixon, L. J. B., Paslaru, E., Hartmann, J., & Jarrar, M.: Identification of Standards on Metadata for Ontologies. KWeb Deliverable D1.3.2 (2005) Wellisch, H. H.: Glossary of Terminology in Abstracting, Classification, Indexing, and Thesaurus Construction (2nd ed., 2nd pr). Medford, NJ: Information Today (2003) Zeng, M.L.: Taxonomy of knowledge organization sources / systems. Online: http://nkos.slis.kent.edu/kos_taxonomy.htm [January 10, 2008] (2000) 30
Websites CAS Chemical Abstract Services: http://www.cas.org/ DAML Ontologies: http://www.daml.org/ontologies/ DDC Dewey Decimal Classification: http://www.oclc.org/dewey/ Del.icio.us: http://del.icio.us Flickr: http://www.flickr.com Gene Ontology (GO): http://www.geneontology.org/ IPC International Patent Classification: http://www.wipo.int/classifications/ipc/ipc8/?lang=en MeSH Medical Subject Headings: http://www.nlm.nih.gov/mesh/ OBO: http://www.obofoundry.org/ ODP Open Directory Project: http://www.dmoz.org OMV Ontology Metadata Vocabulary: http://omv.ontoware.org/ SchemaWeb: http://www.schemaweb.info/schema/browseschema.aspx Swoogle: http://swoogle.umbc.edu/ WordNet: http://wordnet.princeton.edu/ WortSchatz: http://wortschatz.uni-leipzig.de/ 31