How Matchable Are Four Thousand Ontologies on the Semantic Web

Abstract

A growing number of ontologies have been published on the Semantic Web by various parties, to be shared for describing things. Because of the decentralized nature of the Web, there often exist different but similar ontologies from overlapped domains, or even within the same domain. In this paper, we collect more than four thousand ontologies and perform a large-scale pairwise matching based on an ontology matching tool. We create about three million mappings between the terms (classes and properties) in these ontologies, and construct a complex term mapping graph with terms as nodes and mappings as edges. We analyze the macroscopic properties of the term mapping graph as well as the derived ontology mapping graph, which characterize the global ontology matchability in several aspects, including the degree distribution, connectivity and reachability. We further establish a pay-level-domain mapping graph to understand the common interests between different ontology publishers. Additionally, we publish the generated mappings online based on the R2R mapping framework. These mappings and our observations are believed to be useful for the Linked Data community in ontology creation, integration and maintenance.