Finding Good URLs: Aligning Entities in Knowledge Bases with Public Web Document Representations
Download as PPTX, PDF0 likes342 views
This document summarizes a workshop on aligning entities in knowledge bases with representations on the public web. It presents an experimental evaluation of using label search, exploiting link structure, and type filtering to map 100 entities from knowledge bases to URLs on the public web. The best performing methods were found to be label search and focused HITS, and adding type filtering improved results for all methods. Next steps include further investigating domain-dependent performance.
Finding Good URLs: Aligning Entities in Knowledge Bases with Public Web Document Representations
- 1. Institute for Web Science & Technologies – WeST Workshop Web of Linked Entities (WoLE 2012) at ISWC 2012 Sunday, 11 November 2012 Finding Good URLs: Aligning Entities in Knowledge Bases with Public Web Document Representations Christian Hachenberg and Thomas Gottron
- 2. Mapping Documents to Entities dbpedia.org:Rob_Roy_(film) Finding Good URLs Thomas Gottron WoLE Workshop 2012 2
- 3. Mapping Entities to Documents dbpedia.org:Rob_Roy_(film) Align entities in KB with public documents • Publish knowledge base • Propagate changes • Human readable representation Finding Good URLs Thomas Gottron WoLE Workshop 2012 3
- 4. Task Definition George Lucas type: director dbpedia:George_Lucas type: movie ??? dbpedia:Star_Wars_Episode_IV:_A_New_Hope Star Wars IV: A New Hope 3 types of information: dbpedia:Harrison_Ford • Labels type: actor • Link structure Harrison Ford • Types Finding Good URLs Thomas Gottron WoLE Workshop 2012 4
- 5. Label Search (using Web Search Engine) George Lucas type: director SW4 dbpedia:George_Lucas type: movie SW4 dbpedia:Star_Wars_Episode_IV:_A_New_Hope Star Wars IV: A New Hope SW4 dbpedia:Harrison_Ford Implementation: type: actor Harrison Ford • Bing Finding Good URLs Thomas Gottron WoLE Workshop 2012 5
- 6. Exploiting Link Structure George Lucas type: director GL SW4 dbpedia:George_Lucas type: movie SW4 dbpedia:Star_Wars_Episode_IV:_A_New_Hope Star Wars IV: A New Hope HF Implementation: SW4 dbpedia:Harrison_Ford • In-degree type: actor Harrison Ford • PageRank • HITS Finding Good URLs Thomas Gottron WoLE Workshop 2012 6
- 7. Type Filtering Gran Torino type: movie SW4 dbpedia:Gran_Torino_(film) GT type: movie SW4 dbpedia:Star_Wars_Episode_IV:_A_New_Hope RR Star Wars IV: A New Hope Rob Roy type: movie SW4 dbpedia:Rob_Roy_(film) Implementation: • Borda Count for domain ranking Finding Good URLs Thomas Gottron WoLE Workshop 2012 7
- 8. Experimental Setup 100 Entities 4 domains (cities, companies, persons, movies) Stratified by little, medium and large representation on the web Complete network of linked entities Application of label search and link structure approaches Type-filtering as post-process User evaluation (Cranfield setup, pooling) Graded relevance judgements High juror agreement (Krippendorff's Alpha >0.67) Finding Good URLs Thomas Gottron WoLE Workshop 2012 8
- 9. Evaluation Metrics Finding Good URLs Thomas Gottron WoLE Workshop 2012 9
- 10. Evaluation: Results Statistically significant , p=0.05 Finding Good URLs Thomas Gottron WoLE Workshop 2012 10
- 11. Evaluation: Results (Domain, Stratum) Finding Good URLs Thomas Gottron WoLE Workshop 2012 11
- 12. Evaluation: Results (Filtering) Finding Good URLs Thomas Gottron WoLE Workshop 2012 12
- 13. Conclusions and Next Steps Novel task: Mapping entities to public web URLs – Evaluated 9 link analysis and web search methods (+1 post- processing using Borda counts) – Best methods: Label Search and Focussed HITS • Semantic Typing boosts all results Next steps: Investigate domain-dependent performance of methods Finding Good URLs Thomas Gottron WoLE Workshop 2012 13
- 14. Thank you! Contact: WeST – Institute for Web Science and Technologies Universität Koblenz-Landau gottron@uni-koblenz.de Finding Good URLs Thomas Gottron WoLE Workshop 2012 14