HOM: An approach to calculating semantic similarity utilizing relations between ontologies

Abstract

In the Internet environment, ontology heterogeneity is inevitable due to many coexistent ontologies. Ontology alignment is a popular approach to resolve ontology heterogeneity. Ontology alignment establishes the relation between entities by computing their semantic similarities using local or/and non-local contexts of entities. Besides local and non-local context of entities, the relations between two ontologies are helpful for computing their semantic similarity in many situations. The aim of this article is to improve the performance of ontology alignment by using these relations in similarity computing. A hierarchical Ontology Model (HOM) which describes these relations formally is proposed followed by HOM-Matching, an algorithm based on HOM. It makes use of the relations between ontologies to compute semantic similarity. Two groups of experiments are conducted for algorithm validation and parameters optimization. © 2008 Springer-Verlag Berlin Heidelberg.