Matching biomedical ontologies through Compact Differential Evolution algorithm with compact adaption schemes on control parameters

Biomedical ontology is a unified model for describing biomedical knowledge, which can be of help to solve the issues of heterogeneity in different biomedical databases. However, the existing biomedical ontologies could define the same biomedical concept in different ways, which yields the biomedical ontology heterogeneous problem. To implement the inter-operability among the biomedical ontologies, it is critical to establish the semantic links between heterogenous biomedical concepts, so-called biomed-ical ontology matching. Evolution Algorithm (EA) is a state-of-the-art methodology for matching ontolo-gies, but two main shortcomings, i.e. the huge memory consumption and long runtime, make it incapable of effectively matching biomedical ontologies. In this work, a novel Adaptive Compact Differential Evolution algorithm (ACDE) is proposed to solve the biomedical ontology matching problem, which uti-lizes a compact encoding mechanism to save the memory consumption and introduces the compact adaption schemes on control parameters to improve the algorithm's converging speed. The experiment exploits four biomedical ontology matching tracks, which are provided by the famous Ontology Alignment Evaluation Initiative (OAEI), to test ACDE's performance. The experimental results show that ACDE can effectively reduce EA-based ontology matcher's memory consumption and runtime, and its results significantly outperform other EA-based matchers and OAEI's participants. (c) 2020 Published by Elsevier B.V.