SemRep: A Repository for Semantic Mapping

In schema and ontology matching, background knowledge such as dictionaries and thesauri can considerably improve the mapping quality. Such knowledge resources are especially valuable to determine the semantic relation type (e.g., equal, is-a or part-of) that holds between related concepts. Previous match tools mostly use WordNet as their primary resource for background knowledge, althoughWordNet provides only a limited coverage and currentness. We present the design and use of a new comprehensive repository called SemRep that combines concepts and semantic relationships from different resources. It integrates both manually developed resources (including WordNet) and semi-automatically extracted relations from Wikipedia. To determine the semantic relationship between two concepts of interest, SemRep also considers indirect relationships of possibly different types. An initial evaluation shows the general effectiveness and efficiency of using SemRep for ontology matching.