Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

EDF2012 Mariana Damova - Factforge


Published on

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

  • Be the first to like this

EDF2012 Mariana Damova - Factforge

  1. 1. FactForge  FactForgeData Service and the Value of  Inferred Knowledge Mariana Damova, PhD European Open Data Forum June 2012                             
  2. 2. Ontotext – Top‐5 provider of core Semantic Technology – Established in year 2000; offices in Bulgaria, UK, USA – Active both in research and commercial projects p j• 360° semantic technology – unique portfolio: – Semantic Databases: high‐performance RDF DBMS, scalable reasoning – Semantic Search: text‐mining (IE), metadata generation, Information Retrieval (IR) – Web Mining: focused crawling, screen scraping, data fusion  – Linked Data Management and Data Integration Good recognition in the SemTech community – Ontotext pages are ranked #1 for “semantic annotation” and “semantic repository” at  Ontotext pages are ranked #1 for  semantic annotation and semantic repository at GYM, #3 for “linked data management” at Google Several joint ventures and subsidiaries – Innovantage: leading online recruitment intelligence provider in UK
  3. 3. Ontotext Clients (selected) British Broadcasting Corporation (BBC) – Run its World Cup 2010 sites on top of OWLIM – Since Mar’12 BBC Sports and 2012 Olympics sections are driven Since Mar 12 BBC Sports and 2012 Olympics sections are driven  by OWLIM and a Concept Extraction service developed by Ontotext Press Association (UK) – Analysis of Sports news Analysis of Sports news – Concept extraction – Linked data generation Top‐3 USA media (not allowed to name) Top‐3 USA media (not allowed to name) The National Archives (UK) contracted Ontotext to implement  semantic KB and semantic search for the Government Web Archive British Museum (UK) Ontotext leads the development of Phase 3 of  ResearchSpace project on collaborative research in cultural heritage;  British Museum’s public SPARQL end‐point is powered by OWLIM de Bibliothek (Holland)  d Bibli h k (H ll d) aggregation of data from 150 library databases
  4. 4. Linked Open Data is maturingLinked Open Data is maturing LOD cloud grows by billions of triples yearlyTechnologies and guidelines about T h l i d id li b t how to produce linked data fast  how to assure their quality  h h l how to provide vertical oriented data services LOD2, LATC, baseKB  European Data Forum June2012 #4
  5. 5. This talk is about This talk is about reasoning  and  d coping with diversity of the data on the web of data  European Data Forum June 2012 #5
  6. 6. Outline• FactForge (beta)• Reference Layer• Access Modes• Querying  – Airports around London – US city – a subject of a Novel y j – US city – contactInformation• Challenges• Conclusion European Data Forum
  7. 7. FactForge (beta)the largest body of heterogeneous general knowledge on which inference has been performed g y g g g p– powered by OWLIM 5.0 – supporting SPARQL 1.1 European Data Forum
  8. 8. Datasets REASON‐ABLE VIEW of LOD datasets Number of explicit statements: 1,796,673,630 b f l Implicit statements: 1,3 Retrievable statements:  14,928,925,039 CIA FactBook DBpedia 3.7 Freebase NY Times Lexvo Wordnet 3.0 Geonames Lingvoj MusicBrainz materialization is performed with respect to the semantics of OWL-Horst optimized European Data Forum
  9. 9. Reference Layer PROTON – light weight upper level ontology ~500 classes, ~150 properties , p p at schema level:(1) using rdfs:subClassOf and rdfs:subPropertyOf statements;(2) using OWL expressions where there is a difference in the conceptualization(3) using inference rules if additional individuals are necessary in the repository to support the mapping European Data Forum June 2012 #9
  10. 10. Access modesRDF Search - retrieve ranked list of URIs related to literals, which contain specific keywords European Data Forum June 2012 #10
  11. 11. Access modes (condt) Exploration ‐ traversing the data, one resource at a time 
  12. 12. Access modes (condt) Exploration ‐ traversing the data, one resource at a time,   inspecting inferred knowledge - locatedIn – Denmark, Northern Europe- Geonames types/FearureCodes P.PPL yp /- parentFeature – Denmeark, Europe… European Data Forum June 2012 #12
  13. 13. Access modes (condt) Exploration - traversing the data, one resource at a time, inspecting inferred knowledge- locatedIn - Europe- subRegionOf - Europe g p- hasContactInfo – website via Freebase- containsLocation … European Data Forum June 2012 #13
  14. 14. Access modes (condt)SPARQL endpoint European Data Forum June 2012 #14
  15. 15. Access modes (condt)RelFinder European Data Forum June 2012 #15
  16. 16. QueryingUsing LOD concepts SELECT * WHERE { ?Person dbp‐ont:birthPlace ?BirthPlace ; rdf:type dbp‐ont:Politician ; ?BirthPlace geo‐ont:parentFeature dbpedia:Germany . ?BirthPlace geo ont:parentFeature dbpedia:Germany } Using the intermediary layer g y y SELECT * WHERE { ?Person prot:birthPlace ?BirthPlace ; rdf:type prot:Politicianr ; ?BirthPlace prot:subRegionOf dbpedia:Germany . } European Data Forum June 2012
  17. 17. Find Airports near London Standard LOD vs. PROTON query 13 vs. 20 results DBpedia vs DBpedia and Geonames vs. European Data Forum June 2012 #17
  18. 18. Find airports near London ‐ Results comparison Using Geospatial index of OWLIM g p European Data Forum June 2012 #18
  19. 19. City – a subject of a science fiction author European Data Forum June 2012 #19
  20. 20. OWLIM 5.0 and SPARQL 1.1Exemplary queries :GROUP BY, minGROUP BY, min — Minimal and maximal population counts of European countriesFederated Query between FactForge and LinkedLifeData —D Drugs that cure the disease from which died Alexandre Graham Bell th t th di f hi h di d Al d G h B llLiteral index over dates – World governors in office between 1980 and 2005Literal index over digits ― European countries with population above 20 MLNGeospatial indexGeospatial index — Show the distance from London of airports located at most 50 miles away from it European Data Forum June 2012 #20
  21. 21. Challenges and usage• Clean data  – Clean up input data• At model level  – Contradiction detection – Consistency checking y g• Curation and upgrading methodology FactForge has been used as data layer infrastructure in FP7 projects, like RENDER FactForge has been used in tasks of linked data generation from unstructured data, metadata enrichment of structured data providing linkages to the entire LOD cloud for example The National Archive European Data Forum June 2012 #21
  22. 22. Acknowledgements Partial fundingColleaguesAtanas Kiryakov, CEO of OntotextZdravko T hZd k Tashev, Ontotext OIvan Peikov, OntotextRouslan Velkov, OntotextKiril Simov, OntotextBarry Bi hB Bishop, Ontotext OBarry Norton, OntotextMarin Dimitrov, OntotextAlex Simov, OntotextJordan Di hJ d Dichev, Ontotext O t t t k LinksKonstantin Penchev, Ontotext Email: E il European Data Forum June 2012 #22
  23. 23. Thank you for your attention!