The problem of identifying semantically aligned entities in different ontologies known as ontology mapping is an outstanding research area and lies at the heart of many semantic applications. The overarching goal of ontology mapping is to discover a valid and comprehensive mapping with the aim of maximizing the number of reasonable alignments of ontological entities. Recently many efforts to automate the ontology mapping have been carried out, with some problems such as scalability and efficiency still evident. In this paper, ontology mapping in heterogeneous knowledge bases is formalized as an optimization problem, and an efficient method called Harmony Search based Ontology Mapping (HSOMap) is proposed, that effectively finds a near-optimal mapping for two input ontologies. In this approach, we make use of many kinds of rating functions, which are also called base matchers to evaluate the similarity of entities. Each base matcher captures the similarity between entities from a different perspective and is able to exploit the available side information about the entities effectively. Also, a novel weighted harmonic-mean method is proposed to aggregate different metrics into a single similarity metric among all pairs of entities from two ontologies. After obtaining the combined similarity metric between ontological entities, a discrete harmony search algorithm is proposed to extract the best alignment. To demonstrate the merits and advantages of the HSOMap algorithm, we conduct a set of experiments on benchmark data sets and compare its performance to other state-of-the-other methods. Our experimental results demonstrate that applying harmony search in the context of ontology mapping is a feasible approach and improves the mapping effectiveness significantly. (C) 2016 Elsevier Inc. All rights reserved.
展开▼