An Evolutionary Tabu Search Algorithm for Matching Biomedical Ontologies

Since these biomedical ontologies are mostly developed independently and many of them cover overlapping domains, establishing meaningful links between them, so-called biomedical ontology matching, is critical to ensure inter-operability and has the potential to unlock biomedical knowledge by bridging related data. Due to the complexity of the biomedical ontology matching problem (large-scale optimal problem with lots of local optimal solutions), Evolutionary Algorithm (EA) can present a good methodology for determining biomedical ontology alignments. However, the slow convergence and premature convergence are two main shortcomings of EA-based ontology matching techniques, which make them incapable of effectively searching the optimal solution for biomedical ontology matching problems. To overcome this drawback, in this paper, an Evolutionary Tabu Search Algorithm (ETSA) is proposed, which introduces the Tabu Search algorithm (TS) as a local search strategy into EA's evolving process. Moreover, to efficiently solve the biomedical ontology matching problem, an biomedical concept similarity measure is presented to calculate the similarity value of two biomedical concepts and an optimal model for biomedical ontology matching is constructed. The experiment is conducted on the Large Biomed track provided by the Ontology Alignment Evaluation Initiative (OAEI), and the comparisons with state-of-the-art ontology matchers show the effectiveness of ETSA.