Taking advantages of ontology and contexts to determine similarity of data

Abstract

Data integration is the process of unifying data sharing some common semantics but are originated from unrelated sources. In our work we consider these sources are autonomous, heterogeneous and they are physically distributed. These three characteristics make the integration task more difficult as there are several aspects to bear in mind. In this work we only focus on one of these aspects, the semantic heterogeneity, which deals with the meaning of the concepts within the information sources. As each source contains a specific vocabulary according to its understanding of the world, terms denoting same meaning can be very difficult to find. In this paper we will briefly explain our method to find similarities using ontologies and contexts. We will propose some improvements in the similarity functions in order to take advantages of the information the ontologies provide.