Automatic selection of background knowledge for ontology matching

Abstract

Background knowledge in form of ontologies is an important source of information for many tasks in the semantic web, e.g., ontology matching, ontology construction and editing, natural language processing. In particular, ontology matching and integration can benefit from background ontologies as semantic relationships may be discovered which cannot be identified otherwise. In existing approaches, the background ontology has to be provided often by the user. Therefore, we present an approach that uses background knowledge for matching; but in contrast to other approaches, our approach is able to identify appropriate background ontologies automatically. We implemented this approach in our matching frame-work GeRoMeSuite and tested it with several data sets from the Ontology Alignment Evaluation Initiative (OAEI) campaign. The evaluation shows that the use of background knowledge improves the result in most cases and that our ontology discovery process is able to find appropriate background knowledge to bridge the gap between the two input ontologies.