Linked Data Integration


Published on

Présentation at Stanford logic lab in August 2009 about data integration and linked-data

Published in: Technology, Education
1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Linked Data Integration

  1. 1. Structured data interoperability on the Web Fran¸ois Scharffe c INRIA Grenoble Rhone-Alpes, France August 19, 2009 Stanford Logic Group
  2. 2. Introduction The Web of data/Semantic Web is the future of structured data ! I’ll talk about. . . structured data convergence of structured data extraction and structured data publication (SW and linked-data) Interoperability in the Web of data: issues, research and achievments I’ll suppose that. . . You have a relational database approach to structured data You know about the Web You’ve heard about the semantic Web: RDF, OWL, SPARQL
  3. 3. Structured data and the Web Structured data is the future of the Web Many ongoing research aiming to extract structured data from the Web Deep Web, spreadsheets, forms Fusion tables Effort to provide formalisms, standards and tools to publish structured data and vocabularies: Semantic Web and linked-data RDF, OWL, RDFa, RIF Semantic Wiki
  4. 4. The web of Data and the open linked data project Publish vocabularies useful Web ontologies Publish data RDF dereferenceable URIs SPARQL endpoints Link to existing data
  5. 5. Linked data cloud 2009-03- 05 colored.png
  6. 6. Interoperability issues Vocabs Vocabularies should be reused, but anyone is free to provide its own Any published data sources might use its own schema as an ontology Results in many ontologies/vocabularies available and the need to connect them Data Even when a same ontology is used, if two data-sources are published, they might contain equivalent resources Need to interconnect these resources
  7. 7. Interconnecting vocabularies Ontology matching and alignment is the key to issue one: Matching algorithms GUIs alignment representation patterns mediators Results: Alignment API and server, Ontomap, OAEI, Alignment patterns language Data linkage is the key to issue 2 Matching algorithms Link specification languages Linksets Results: Silk, Knofuss, RDF-AI, VOID
  8. 8. Linking Data Class 4 Class 3 Class 1 Class 2 Class 5 rdf:type rdf:type rdf:type rdf:type rdf:type rdf:type rdf:type G1 G2 r1-1 r2-5 r2-1 r1-4 r1-3 r2-2 r1-2
  9. 9. Linking Data G1 A G2 c1 r1-1 r2-5 r2-1 r1-4 r1-3 r2-2 r1-2 c2
  10. 10. Linking Data Graphs Alignment Fused Graph Linkset Graphs Graph Alignment Graph Fused Graph Parameters Parameters Parameters Parameters Linkset Output Output Output Output Input Input Input Input Fusion Pre-processing Matching Post-processing Interlink RDF file SPARQL Endpoint
  11. 11. Aligning Vocabularies Class Class Attribute Relation Relation Correspondence Class Attribute Attribute Class Correspondence Val1 Att1 Value Val2 Att2 Transformation Concatenation Val3 Att3 RCA-A concat Pattern Value http://.../francois vc:VCard foaf:Person http://.../francois Attribute Relation vc:name vc:name Correspondence Class Attribute _:bn01 foaf:name foaf:name vc:N Correspondence "Scharffe" vc:family-name Value "François" Transformation value vc:given-name "François Vincent Concatenation Alfred Scharffe" "Vincent Alfred" vc:additional-name RCA-A concat Pattern
  12. 12. Alignment representation vcard:family-name a align:Property; align:bind-with "?x". vcard:given-name a align:Property; align:bind-with "?y". vcard:additional-name a align:Property; align:bind-with "?z". foaf:name a align:Property. :property-group a align:Property; or [ align:item vcard:family-name; align:item vcard:given-name; align:item vcard:additional-name. ]. align:transf [ align:transf-id transf:concat; align:transf-param "?x ?y ?z". ]. :foaf-vcard-names a align:Cell; align:entity1 :property-group; align:entity2 foaf:name.
  13. 13. Using alignments Matcher produces FOAF Alignment vCard in us d- used-in Design Time ed use -in Run Time FOAF Instance Mediated SW Pipe to be mediated vCard Instance uses SPARQL++ Reasoner
  14. 14. Conclusion The data representation is there Data integration works through owl:sameAs Vocabularies integration works though there is no standard representation Where is the killer app ?