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.

Thesaurus based Enterprise Search

2,352 views

Published on

Published in: Technology
  • Be the first to comment

Thesaurus based Enterprise Search

  1. 1. Thesaurus based Enterprise Search <br />Two Show Cases<br />Andreas Blumauer<br />Graz, September 2011<br />
  2. 2. Agenda<br />Semantic searchscenarios<br />The role of thesauri in semantic search<br />PoolParty Semantic Search<br />Live Demo – http://bit.ly/semantic_search<br />Show Cases & Demos<br />2<br />
  3. 3. 3<br />Semantic searchscenarios<br />Semantic searchhasmanyfaces<br />
  4. 4. Situations in which semantic search can help<br />4<br />I want to seefactsfrom different sourcesdescribingthisentity.<br />I can´trememberhow to spell the searchterm<br />I want to knowmoreaboutthisentity in a certaincontext.<br />I want to search in different languagessimultaneously<br />I want to gainbackgroundknowledge to a certaindocument<br />I want the software to understand what I meanby „Jaguar“<br />I can´trememberexactlywhat I was lookingfor<br />
  5. 5. Knowledge worker´squestions<br />5<br />What is state of <br />the art in xy?<br />Hasanybodysolved<br />the problemxy?<br />Who can I ask<br />aboutxy?<br />Whatare the otherscurrentlyworking on?<br />
  6. 6. Four demands for a smarter search<br />Find information faster Provide search assistants<br />Reveal hidden information Enrich the search index with background knowledge<br />Find more specific informationQuery the semantic web<br />Findlinked informationIntegrate data sources<br />6<br />
  7. 7. Find information faster – Auto-Complete<br />7<br />I can´trememberhow to spell the searchterm<br />To provide powerful auto-complete also forenterprisesearch<br />scenariosyouneed to establish an enterprisevocabulary.<br />
  8. 8. Find information faster – Status quo<br />8<br />I can´trememberexactlywhat I was lookingfor<br />hydropower plants<br />Search<br />Small hydro<br />Search<br />
  9. 9. Find information faster with related search terms<br />9<br />hydropower plants<br />Search<br />http://www.reegle.info/clean-energy-search<br />
  10. 10. Reveal hidden information – Status quo<br />10<br />I forgotsome of the namesfor the entityI´mlookingfor<br />SNCR<br />Search<br />SNCR OR „Selective non-<br />Search<br />
  11. 11. Reveal hidden information with query expansion<br />11<br />SNCR<br />Search<br />OR "selective non catalytic reduction"<br />SNCR<br />preferred Label<br />alternative Label<br />selective non <br />catalytic reduction<br />
  12. 12. Multi-lingual search based on a thesaurus<br />12<br />I want to search in different languagessimultaneously<br />clean energy<br />Search<br />OR energíalimpia<br />preferred Label @en<br />clean energy<br />energíalimpia<br />preferredLabel @es<br />
  13. 13. Reveal hidden information and relations<br />I want to gainbackgroundknowledge to a certaindocument<br />Find documents<br />or images related<br />to any other text.<br />http://poolparty.punkt.at/demozone<br />13<br />
  14. 14. Find more specific information – Status quo<br />14<br />I want to knowmoreaboutthisentity in a certaincontext.<br />Goldman Sachs<br />Search<br />3 different contextsfor<br />„Goldman Sachs“:<br /><ul><li> Bond issuer
  15. 15. Analyst
  16. 16. Stock</li></li></ul><li>Find more specific information with faceted search<br />15<br />Zero-resultqueries<br />won´t happen anymore<br />facetssupport<br />structuredqueries<br />facetshelp<br />to drill down <br />searchresults,<br />adaptdynamically<br />
  17. 17. Complex queries with faceted search over linked data<br />16<br />„Show me all airlineswhoseparentcompany is Lufthansa“ <br />http://dbpedia.neofonie.de/<br />
  18. 18. MyEnergy-Dossier about<br />Find linked information – Status quo<br />17<br />I want to seefactsfrom different sourcesdescribingthisentity.<br />The userhas to put<br />togethermanuallyenergy-related<br />informationabouta country.<br />
  19. 19. 360O views: Find linked information<br />18<br />Energy-related<br />informationabout countries<br />are „mashed“ automatically<br />byusing „linkeddata“<br />http://www.reegle.info/countries<br />
  20. 20. Add personal context to the search<br />19<br />Jaguar<br />Search<br />I want the software to understand what I meanby „Jaguar“<br />
  21. 21. 20<br />The role of thesauri in semantic search<br />How verticalsearchcanbenefitfromknowledgemodels<br />
  22. 22. The role of thesauri in semantic search<br />21<br />
  23. 23. The role of thesauri in semantic search (contd.)<br />22<br />Thesaurus as the centralpoint<br />to control:<br /><ul><li>labels & queryexpansion
  24. 24. facets
  25. 25. refinesearchmechanisms
  26. 26. metadataintegration</li></li></ul><li>Data integration and schema mapping based on thesauri<br />23<br /><person><br /> Thomas Miller<br /></person><br /><employee><br /> Tom Miller<br /></employee><br />Source 1<br />Source 2<br />
  27. 27. Usage of linkeddatafor semantic search<br />Alignthesaurusconceptswith DBpedia resources<br />disambiguation!<br />performance!<br />Enrichconceptwithcategoryinformation<br />schema.org / DBpedia ontology<br />YAGO/Umbel<br />Use categoryinformationforconcepts<br />to categorizedocument (usage of transitivities)<br />to providesearchfacets<br />24<br />
  28. 28. 25<br />PoolParty Semantic Search (PPS)<br />Make semantic searchcometrue!<br />
  29. 29. PoolParty System Architecture<br />26<br />SearchApplication<br />Search Services<br />SemanticIndexer<br />Collector<br />Document Index<br /><xml><br />Cartridge<br />
  30. 30. Indexing and Mapping with PoolParty<br />Metadata Standards<br />Rich metadata in a standardized, extensible format (SKOS / RDF)<br />Document metadata is mapped to concepts in the thesaurus<br />Cost efficient metadata management<br />Thesaurus is managed with PoolParty´s easy-to-use Thesaurus Manager<br />One central metadata repository<br />Improved end-user experience<br />Semantic information improves search experience<br />27<br />
  31. 31. PoolParty Search API & Standard GUI<br />28<br /><ul><li>Available web services:
  32. 32. Search Service
  33. 33. Suggest Service
  34. 34. Similarity Service
  35. 35. Supportedformats:
  36. 36. JSON
  37. 37. XML
  38. 38. RSS</li></ul>http://bit.ly/semantic_search<br />
  39. 39. PoolParty Semantic Search Demo – Result<br />29<br />specifyyourquery<br />withcategorised<br />auto-complete<br />storequeries<br />withsearchbasket<br />select proper<br />facets<br />find similardocumentsforrelevant results<br />facetssupport<br />structuredqueries<br />http://bit.ly/semantic_search<br />
  40. 40. 30<br />Show cases & Demos<br />Thesaurus basedsearch on the web & intranet<br />
  41. 41. Show Case No. 1: Semantic Search based on reeglethesaurus<br />31<br />Web catalogue<br />of actors<br />SearchApplication<br />Search Services<br />Projects DB<br />SemanticIndexer<br />Collector<br />Document Index<br />Actors DB<br />Cartridge<br />Thesaurus<br />31<br />
  42. 42. Data integration based on Reegle thesaurus<br />32<br /><sector><br /> Hydro Power small scale<br /></sector><br /><category><br /> Micro Hydro<br /></category><br />Web catalogue<br />Actors DB<br />
  43. 43. Show caseNo. 2 - www.reegle.info<br />33<br />
  44. 44. Show Case No. 3: Very large financialinstitute<br />34<br />SearchApplication<br />Search Services<br />DMS 1<br />SemanticIndexer<br />Collector<br />Document Index<br />DMS 2<br />Cartridge<br />VLFIThesaurus<br />34<br />
  45. 45. Contact<br />Andreas BlumauerManaging Partner, CEOa.blumauer@semantic-web.at<br />Alexander KreiserSystem Architecta.kreiser@semantic-web.at<br />35<br />Semantic Web Company GmbHMariahilfer Straße 70A—1070 Wien / Austria<br />+43-1-4021235 http://www.semantic-web.at/http://www.poolparty.biz/<br />http://bit.ly/semantic_search<br />http://lod2.eu/<br />http://twitter.com/semwebcompany<br />http://linkd.in/oFFnO4<br />

×