Using multiple ontologies as background knowledge in ontology matching

Abstract

Using ontology as a background knowledge in ontology matching is being actively investigated. Recently the idea attracted attention because of the growing number of available ontologies, which in turn opens up new opportunities, and reduces the problem of finding candidate background knowledge. Particularly interesting is the approach of using multiple ontologies as background knowledge, which we explore in this paper. We report on an experimental study conducted using real-life ontologies published online. The first contribution of this paper is an exploration about how the matching performance behaves when multiple background ontologies are used cumulatively. As a second contribution, we analyze the impact that different types of background ontologies have to the matching performance. With respect to the precision and recall, more background knowledge monotonically increases the recall, while the precision depends on the quality of the added background ontology, with high quality tending to increase, and the low quality tending to decrease the precision.