Terminology services and the DDC the High-Level Thesaurus

Terminology services and the DDC: the High-Level Thesaurus and beyond Presented to the symposium Dewey goes Europe: on the use and development of the Dewey Decimal Classification (DDC) in European libraries 28 April 2009, Austrian National Library, Vienna Gordon Dunsire

Overview ² Terminology services ² HILT ² To the Semantic Web and beyond

Terminology services ² “Web services involving various types of knowledge organization resources [vocabularies], including authority files, subject heading systems, thesauri, Web taxonomies, and classification schemes” ²Vizine-Goetz and others (OCLC), 2004 ²Example: Mappings from a term in one vocabulary to one or more terms in another vocabulary

HILT: High-Level Thesaurus project ² Started in 2000. Now in 4 th phase, embedding functionality in pilot services ² Funded by Joint Information Systems Committee, and supported by OCLC ² To provide subject interoperability in a multischeme environment via inter-scheme mapping ² Ideally by identifying a generic approach that can be developed through distributed collaborative action …

Role of DDC in HILT ² DDC used as a semantic translator between different Anglophone subject terminologies ² Availability of DDC mappings to captions, relative index and LCSH ² DDC (partially) mapped to other terminologies ²AAT, Unesco, Me. SH, etc. + some non-English ² DDC also used to classify collections ²“Subject landscaping” gives best places (catalogues) to search for items ²Drill-up: (Item > Collection) = (Specific > General) = Truncation of DDC notation

HILT applications: Disambiguation ² User-supplied subject term may be ambiguous ²E. g. “tree” – genealogy or forestry? ² So, display all subject headings or captions containing the term, along with context ²Hierarchy of “broader” headings ²Driven by DDC notation ²Non-HILT possibilities include definitions, scope notes, etc. ² User chooses heading/caption ²System knows mapped DDC notation

HILT applications: Spell-checking ² User-supplied subject term may have incorrect spelling, or variant spelling ²E. g. “colour” vs “color” ² So, display terms from all HILT vocabularies which pattern-match the user term ² User is alerted if term is erroneous, or variant spellings exist, or term is not present in any vocabulary

HILT applications: Item-level retrieval ² System knows what subject heading or classification scheme is indexed in a specific catalogue or other item-level finding-aid ² User-supplied subject term is matched to semantically-equivalent term in the scheme ² Matched term is used to search finding-aid ²Manually, by user ²Transparently by service, using machine-tomachine interface for finding-aid, if available

HILT applications: Landscaping ² “Landscape” is a set of collections and associated finding-aids which are strong in a specific subject area ² Classify collection-level descriptions with DDC ²More than one notation may be required ² Use implicit hierarchy of decimal notation to identify collections containing a user-supplied term ²User term: teeth = 611. 314 ² 611 = Anatomy museum collection

Putting it all together ² User supplies term ² Term is disambiguated (and spell-checked) ² Collections with strength in subject area of term are identified ² Collection finding-aids are searched with semantically-matched terms appropriate to their specific subject scheme ² User gets item-level metadata from multiple collections

Example collections service: SCONE ² Scottish Collections Network ²Metadata for collections located in Scotland ²Archives, libraries, museums, etc. ² Database drives a set of services collectively known as “Scotland’s Information Landscape” ² Collections with specific subject focus are classified with DDC ²Multiple notations used if necessary ² DDC summary captions used to display a hierarchical, browsable tag cloud

Applying HILT to SCONE ² Replace hard-wired DDC summary captions (tags) with captions from HILT ²Problem: HILT assumes minimum 3 -digit notation, and currently cannot cope with 3 xx, 33 x, etc. ² Add disambiguation and landscaping facilities ² Add suggested terms for searching specific finding-aids ²SCONE already stores metadata about subject schemes ² Intute service (project partner) pilot available

HILT embedded in Intute: http: //www. intute. ac. uk/search_hilt. html

Beyond HILT ² HILT approach too expensive to scale across all subject schemes ²Hence “High level” ² Might be complemented by other methods ²One-one vocabulary mappings ²Associative clusters and mappings ²Similar to DDC-LCSH in Web. Dewey ² Wider context of the Semantic Web ²SKOS vocabularies ²RDFS/OWL ontologies and schemas

Behind HILT ² DDC licensing issues are a barrier to full exploitation of HILT ²And other terminology services ²Including OCLC TS pilot? ² Tension between Semantic Web “open-ness” and OCLC/DDC cost recovery ² Creative Commons Attribution. Non. Commercial-Share. Alike license would help ²Or something similar

In front of HILT ² HILT designed as machine-to-machine service ²HILT “end-users” are other online services ²OCLC Terminology Services pilot also intended for technical consumption ² Emerging need for terminology services for non-professionals ²E. g. Simple drop-down term lists; simple thesaurus browse; etc. ²Generic “widgets”, plug-ins, etc. ?

Immediately in front of HILT ² DDC translations! ² HILT infrastructure is independent of language ²Already has non-English (e. g. Welsh) terms ² Should be technically easy to incorporate translations ²Just another DDC-mapped vocabulary ² Full depth not required ²“High” level

Thank you! ² G. dunsire@strath. ac. uk ² HILT ²http: //hilt. cdlr. strath. ac. uk/ ² Scotland’s Information ²http: //www. scotlandsinformation. com/ ² OCLC Terminology Services pilot ²http: //tspilot. oclc. org/resources/