Presented at: 8th Extended Semantic Web Conference (ESWC2011)
An advantage of Semantic Web standards like RDF and OWL is their flexibility in modifying the structure of a knowledge base. To turn this flexibility into a practical advantage, it is of high importance to have tools and methods, which offer similar flexibility in extracting information from a knowledge base. This is closely related to the ability to easily formulate queries over those knowledge bases. We explain benefits and drawbacks of existing techniques in achieving this goal and then present the QTL algorithm, which fills a gap in research and practice. It uses supervised machine learning and allows users to ask queries without knowing the schema of the underlying knowledge base beforehand and without expertise in the SPARQL query language. We then present the AutoSPARQL user interface, which implements an active learning approach on top of QTL. Finally, we present an evaluation based on the SPARQL query log of the DBpedia knowledge base.
Keywords: Active Learning, Query Interfaces, SPARQL, Supervised Learning
Resource URI on the dog food server: http://data.semanticweb.org/conference/eswc/2011/paper/inductive-approaches/6
Explore this resource elsewhere: