Influence of semantic similarity measures on ontology cross-language mappings

Abstract

Cross-language mappings establish relations between ontology concepts defined in different languages. Similarity measures calculate the degree of relatedness between concepts to support matching between two distinct ontologies. Cross-language matching remains an open research issue due to the difficulties in taking advantage of similarity computation. This article investigates the effects of different semantic similarity measures on the identification of cross-language mappings. We carry out experiments exploring real-world biomedical ontology mappings to comprehend the behaviour of computed similarity values. The obtained results indicate the relevance of the domain-related background knowledge in the effectiveness of semantic measures for ontology cross-language alignment.