notes on thoughtlab / athena wp4 november 13, 2009 antoine isaac [email protected]
Post on 15-Jan-2016
214 views
TRANSCRIPT
Towards semantics-enabled search
• Enhance access to Europeana content by semantics
• Exploiting different types of relations– locatedIn, isBornIn, created…
• Making use of inference– Finding work showing London for a query on UK
• Rich descriptions are already there, in metadata!• Requires to make it properly machine-accessible
Goal: semantics in Europeana v1.0
Building a semantic layer to help accessing content
Stefan Gradmann, EDL D2.5
Europeana Thought Lab
http://europeana.eu/portal/thought-lab.html
Semantic autocompletion
Clustering of results
Baseline: matching concepts' label
A "more specific Egypte"??
A "more specific Egypt"?
A concept more specific than the Egypt one
Following other relations
Following other relations - creator
Following other relations - match
Following other relations – death place
Following other relations – death place
Enabling Technologies
• RDF– Uniform format for data– Amenable to sharing and linking
• OWL – Representation of metadata structures– Amenable to inference
• SKOS– Representation of controlled vocabulary– Allows exploitation of legacy knowledge organization
• Simple but precious!• E.g., hierarchical relationships for cluster creation
Where are the challenges?
• Semantic conversion of data– Using appropriate data models– Enriching legacy metadata
• Semantic alignments– Between description ontologies
vra:depicts rdfs:subPropertyOf dc:subject
– Between concepts in controlled vocabulariesiconclass:bird skos:closeMatch ddc:bird
Alignment of semantic references
Where are the challenges?
• Semantic alignment (c'ed)– Find correspondences between large vocabularies– In a multilingual context
Athena WP4
Seems to fit very nicely into that challenge
• SKOS & SKOSification• Semantic alignment:
From Marie-Véronique & Johann, Lund
"The Athena Thesaurus = network of Athena-compliant micro-thesauri with bridges in-between"
• Focus on multilingual resources
What kind of semantic alignment?
• Fundamental goal:– enhancing semantic interoperability of collections– via the KOSs used for describing them
• Several options…
Structural models for interoperability(British Standard BS8723)
1. Unified structure: one KOS
2. Pairwise relations
3. Backbone structure
VocA
VocB
VocC
VocD
VocB
VocC
VocD
VocA
Structural models for interoperability
ThoughtLab "data cloud"• Not really corresponding to best practice
– More like a "web of data" cloud
• But still, a couple of backbone/central nodes– Again, like a "web of data" cloud
• At some point, we have to deal with what is there• Especially if it's much better than nothing!
Goals of Athena WG4?
• Athena Integrated Thesaurus • or Athena Thesaurus Network?
VocA
VocB
VocC
VocD
AthenaThesaurus
VocC
VocD
VocA
Throwing away integrated thesaurus?
• Individual manual mappings can already be exploited– Dumping them in the semantic layer will bring interesting stuff
• Keeping original vocabularies as access points can be an asset
• But a backbone for museum KOSs is likely to bring more– Especially as an umbrella for all those small controlled lists!
• An unified multilingual thesaurus is always extremely precious to have
Throwing away integrated thesaurus?
Thesaurus integration can be used as a driving scenario
• Issue: mapping without application in mind is tricky– What's the "meaning" of a concept?– archeology; netherlands can perfectly be mapped to excavations for translation of book annotations at KB
• Thesaurus integration can provide with mapping criteria– Two concepts are equivalent if we can fit them in the same place
of a semantic network
Wishlist?
• Again, do not forget that intermediate results (individual mappings) can be very precious
• If you produce them as part of the process anyway, there should be a way to export them– As SKOS?
• Problem: ideally, this would require SKOS versions of the individual "micro-thesauri"– Is that planned?
Thanks!