1. BestPortal and BestMapLessons Learned in Lightweight Semantic Access to Court Proceedings Rinke Hoekstra Rinke Hoekstra
2. Overview Open Government Data Bridging the Gap InformationServing BestPortal VocabularyMapping BestMap Case Frames Discussion Rinke Hoekstra
3. Open Government Data Variety of information types Recently: Paradigm shift Public Services Open Data “It’s the data, stupid!” Pro-active information disclosure … open formats, (user) interface, unique ids But… is `just the data’ enough? Rinke Hoekstra
5. BestPortal BEST Project “BATNA Establishment using Semantic Web Technology” Best Alternative to a Negotiated Agreement Unlawful Act (Tort) Improve access to court proceedings Netherlands Council of the Judiciary http://www.rechtspraak.nl 50 thousand verdicts Rinke Hoekstra
6. Step 1: Open Up! Make court proceedings publicly available (http://www.rechtspraak.nl) But: Full text search is not enough Laymen Lawyers Lawyers have their own language: legalese Bridge the gap between common sense and legal knowledge Rinke Hoekstra
7. Step 2 Semantic annotation of meta and data Embedded RDFa Direct publication as RDF Examples: CEN MetaLex http://data.gov.uk http://data-gov.tw.rpi.edu (http://data.gov in RDF) But: these still do not bridge the gap Rinke Hoekstra
8. Step 3: Back to services, then? Knowledge Intensive services Given a ‘commonsense’ description of a case, … provide the legal qualification Examples LKIF Core + HARNESS … original approach of BEST (van Laarschot, 2005) (In case you’re wondering, it could not deal with exceptions) Rinke Hoekstra
9. Problems Too expensive! Requires complete and coherent case descriptions Significant, if not insurmountable modelling effort Maintenance costs … especially in case law Quality assurance Legal theoretical commitment Mappings are definitions Specified using subsumption or equivalence axioms Rinke Hoekstra
10. Best Horsesehoe in 2005 Rinke Hoekstra Case is inferred to be of some type of liability(articles 6:163-6:197 BW)
11. Search (some statistics) “Fingerprints” Vector of weighted terms Article-based Relevance, 0%-40% Based on prototypical cases Relevance, 10%-83% … added some ‘indicative phrases’ Relevance, 47%-83% Concept-based fingerprints Clear vs. Open-Textured Relevance, 58%-83% Rinke Hoekstra
12. So, what next? `Qualification’ of a case is: Too expensive Theoretical overcommitment and Not necessarily useful for search (!) But then… how should the gap be bridged? Rinke Hoekstra
13. Eureka! Given an annotation task: A layman will describe the facts A legal professional will describe the qualification Rinke Hoekstra Mappings should hold between descriptions of court proceedings, not concepts.
14.
15. … and some necessary technical improvementsRinke Hoekstra
17. Vocabularies Cultural Heritage Museums, libraries Huge repositories of (rich) information Annotated using many different vocabularies (knowledge organization systems) Concept-based information retrieval Europeana portal (http://www.europeana.eu) Rinke Hoekstra
19. Why Vocabulary Mapping Integrated access?Vocabulary mapping! Format (XML to RDF) e.g. via XSLT one-way transformation Structure (VCard:Given + VCard:Family to foaf:name) e.g. via SPARQL++, one-way transformation Concepts (foaf:Person to lkif:Person) Simple Knowledge Organization System (SKOS) Lifting existing KOS’s to the Semantic Web Every skos:Concept is an OWL individual Lightweight semantic relations: broader, narrower, and related. Lightweight mapping relations between skos:ConceptSchemes. Rinke Hoekstra
21. Information retrieval perspectivevs. lightweight semantics No many-to-many mappings Mapping only between pairs of concepts Required for re-indexing and search across collections (Isaac et al. 2007) … fundamental issue SKOS concepts and relations are ‘intensional’ What does a mapping then mean? Implicit assumption of extensionality Rinke Hoekstra
22. Extensional View SKOS relations “Resources annotated by some concept should be retrievable via its broader concept.” SKOS mappings “Resources annotated by some concept should be retrievable via the concepts it is mapped to.” … only means to assess quality Rinke Hoekstra
23. BestMap: Requirements Extensional perspective Concepts as annotations on resources Compatibility Integrated with SKOS Hierarchic mappings Exploit skos:broader and skos:narrower Many-to-many mappings Granularity Context determines whether a mapping holds Flexible and Lightweight A mapping is not the definition of a concept Rinke Hoekstra
25. Mappings as OWL Classes A mapping class: Classifies resources annotated using one vocabulary, and Infers annotations using the other vocabulary … it may be directed Rinke Hoekstra
28. BestMap: Discussion Extensible Any OWL axiom may be used in a mapping (e.g. someValuesFrom etc.) Reusable (partial) mappings Exclude resources annotated with a particular concept Negative property assertions Novel “Reification” wrt. normal OWL ontologies Overcomes limitations of SKOS semantics Makes explicit the extensional perspective that underlies SKOS semantics Non-intrusive Rinke Hoekstra
31. … but now the mappings Problem The Good Ol’ Knowledge Acquisition Bottleneck BestMap provides little structure Solution Case Frames (!) Focus on the Unlawful Act Thematic Roles Rinke Hoekstra
32. Quick DEMO (2) BestMap-based Case Frames in BestPortal Rinke Hoekstra
33. Future Work Apply BestMap to other domains Legal assessment based on spatial plans … Further development of BestPortal Case Frames Do the mappings actually work? Does BestPortal really improve access to court proceedings? Connect to the linked data cloud Rinke Hoekstra