DBpedia past, present & future
3 likes1,282 views
1) DBpedia started in 2007 when S旦ren Auer extracted infobox data from Wikipedia pages into RDF triples and collaborated with others to publish the first version. 2) DBpedia has grown significantly since then and the 2014 version contains over 4.5 million entities and 583 million triples extracted from over 100 languages. 3) For DBpedia to continue evolving, areas of focus include fusing data from different sources, validating information, using natural language processing to extract more from Wikipedia text, and developing enterprise solutions to integrate DBpedia knowledge graphs.
DBpedia past, present & future
- 1. past, present & future DBpedia Community Meeting 25.06.15 Poznan
- 3. Get me all soccer players, who played as goalkeeper for a club that has a stadium with more than 40.000 seats and who are born in a country with more than 10 million inhabitants
- 6. How it all started - 2006 - S旦ren Auer (busy with his PhD) asking people: Wikipedia fact tables look like triples, dont you want to write some extractor? - 6 months later: S旦ren wrote the extractor himself and asked Jens Lehmann to help with writing a paper - Chris Bizer : We are extracting people and place information from Wikipedia too lets join efforts and call it DBpedia. - Kingsley Idehen: I need a showcase for my Virtuoso triple store.
- 8. Taking a closer look at heterogeneity - DBpedia Mappings wiki
- 9. Milestones - 2008: DBpedia Live - 2009: Scala-Based framework - 2009: Mappings wiki - 2011: Internationalization - 2011: DBpedia Spotlight - 2014: DBpedia Association (S. Hellmann)
- 10. Now DBpedia 2014 (English): 4.58 mio. entities and 583 mio. triples 131,2 mio. fact assertions (derived from infoboxes) 168,5 mio. triples representing Wikipedia structure 57,1 mio. links to external datasets Localized DBpedia version for 125 languages, built from corresponding Wikipedia versions 12 DBpedia language chapters
- 11. Now X
- 13. DBpedia has to evolve - Fusion - Validation - NLP - Enterprize
- 14. Fusion
- 15. Validation
- 16. NLP - Exploit the text - Let different NLP tools & approaches compete for the best quality (in a certain language) - Need to define the interface (help needed)
- 17. Every Enterprise needs its DBpedia - Represent common sense knowledge (DBpedia and other LOD datasets) as well as the specific enterprise knowledge - Crystallization points for Linked Data intranets an addition to SOA facilitating enterprise-wide data linking & integration - Slicing & Dicing
- 18. Other ideas? add them in http://dbpedia.org/ideas (404)
- 20. The soccer players (for the curious)
- 21. Thank you Big thanks to S旦ren Auer & Markus Ackermann for slide contributions
- 22. Big thanks to S旦ren Auer & Markus Ackermann for slide contributions