Computation of ontology resemblance coefficients for improving semantic interoperability

Abstract

In open and dynamic environments, where various heterogeneous agents need to communicate, a shared ontology that explicitly and formally describes the whole domain of interest, or an alignment that provides semantically related entities among distinct ontologies, can be employed. The former case is infeasible, because a unique conceptual view of a domain is not widely accepted. Hence, the case, usually adopted, is the latter, where independently developed heterogeneous ontologies exist. A challenging issue is how an agent, charged with the task of carrying out the alignment, should select a suitable execution of matchers, to establish correspondences between ontology entities, in the fastest and most efficient way. A solution to this challenge is to use metrics for estimating the resemblance of a given pair of ontologies. To this end, we propose two metrics, as similarity coefficients, to estimate the lexical or structural resemblance of a given pair of ontologies.