Being able to determine and quantify the semantic similarity between ontological concepts is a key requirement for the discovery, selection, and composition of semantic web services. Especially concept similarity is calculated between different service ontologies, it is still a problem. In this paper, we propose an ontology-based approach for semantic service selection which takes into account the heterogeneity of service descriptions. The approach is based on an application ontology (AO), which is merged by different service ontologies and constructed as a semantic net with multiple concept relations. Empirically the proposed method is shown to improve the quality of service selection.