Generierung von Ontologien aus Linked Data

• Antragstellerinnen / Antragsteller: Professor Dr. Jens Lehmann; Dr. Johanna Völker
• Fachliche Zuordnung: Bild- und Sprachverarbeitung, Computergraphik und Visualisierung, Human Computer Interaction, Ubiquitous und Wearable Computing
• Förderung: Förderung von 2013 bis 2016
• Projektkennung: Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 233626375

• Erstellungsjahr: 2015

Zusammenfassung der Projektergebnisse

In this project, we made first steps towards combining the previously distinct logical and lexical ontology learning areas. By extending a formerly pure logic based approach with statistical methods which can be used on text corpora, we were able to foster the generation of more intuitive class expressions. We performed an extensive manual evaluation which showed that the integration of relevance measures can significantly improve results. We see the following topics as relevant for future work based on the GOLD project: A relevant plan for future work is to closely integrate the output of the lexical analysis into the induction process. Due to time restrictions, we did not implement this idea, but believe that it might positively influence the search in the hypotheses space, thus, might result in a faster generation of more intuitive solutions first. Furthermore, one can extend our approach by more sophisticated word sense disambiguation techniques, which will help us to more accurately identify mentions of ontology entities in the text. An improve in accuracy could also be achieved by including WordNet and other lexical resources that can facilitate the detection of words which are synonymous to ontology entity labels.

Projektbezogene Publikationen (Auswahl)

• Pattern Based Knowledge Base Enrichment. In: 12th International Semantic Web Conference, 21-25 October 2013, Sydney, Australia, Springer, 2013
  Lorenz Bühmann, Jens Lehmann
  
• Correlation-based Refinement of Rules with Numerical Attributes. In: Proceedings of the twentyseventh International Conference of the Florida Artificial Intelligence Research Society (FLAIRS): May 21 - 23, 2014 Pensacola Beach, Florida, USA; 345-350. AAAI Press, Palo Alto, Calif., 2014
  André Melo, Martin Theobald and Johanna Völker
  
• Detecting Errors in Numerical Linked Data Using Cross-Checked Outlier Detection. In: Lecture Notes in Computer Science The Semantic Web - ISWC 2014: 13th International Semantic Web Conference, Riva del Garda, Italy, October 19-23, 2014. Proceedings, Part I; 357-372. Springer Internat. Publ., Cham, 201
  Daniel Fleischhacker, Heiko Paulheim, Volha Bryl, Johanna Völker and Christian Bizer
  (Siehe online unter https://doi.org/10.1007/978-3-319-11964-9_23)
• Inductive Lexical Learning of Class Expressions. In: Lecture Notes in Computer Science Knowledge Engineering and Knowledge Management: 19th International Conference, EKAW 2014, Linköping, Sweden, November 24-28, 2014, Proceedings; 42-53. Springer, Berlin [u.a.], 2014
  Lorenz Bühmann, Daniel Fleischhacker, Jens Lehmann, André Melo and Johanna Völker
  (Siehe online unter https://doi.org/10.1007/978-3-319-13704-9_4)
• Perspectives on Ontology Learning. IOS Press / AKA, 2014
  Jens Lehmann and Johanna Völker (eds.)
  
• Repairing Learned Ontologies. WoDOOM 2014: Proceedings of the Third International Workshop on Debugging Ontologies and Ontology Mappings co-located with 11th Extended Semantic Web Conference (ESWC 2014) Anissaras/Hersonissou, Greece, May 26, 2014, 2014
  Daniel Fleischhacker