SBVR Based Natural Language Interface to Ontologies

  • Algirdas Šukys Kaunas University of Technology
  • Lina Nemuraitė Kaunas University of Technology
  • Rita Butkienė Kaunas University of Technology
Keywords: Natural language interface, semantic search, SPARQL, OWL ontology, SBVR business vocabulary, SBVR questions, ATL, Acceleo

Abstract

The semantic search over ontologies allows user to retrieve more relevant results comparing with ordinary keyword based search systems. This type of search system is powered by ontologies and the most convenient interface to ontologies is natural language interface. In this paper, we present multilingual SBVR based natural language interface to ontologies, which allows writing questions based on concepts of SBVR vocabulary and transforms them to SPARQL queries using model transformations. The solution can also be used for questioning, when question mapping to ontology is not straightforward. The experimental evaluation of correctness using Mooney Natural Language Learning Data showed results, similar to other natural language interface solutions, answering questions in English and Lithuanian languages.

DOI: http://dx.doi.org/10.5755/j01.itc.46.1.13998

Author Biographies

Algirdas Šukys, Kaunas University of Technology
Assistant, Department of Information Systems, Faculty of Informatics.
Lina Nemuraitė, Kaunas University of Technology
Professor, Department of Information Systems, Faculty of Informatics.
Rita Butkienė, Kaunas University of Technology
Associate professor, Department of Information Systems, Faculty of Informatics.
Published
2017-03-13
Section
Articles