Index terms: ontology, ontology mapping, system architecture, data integration, genetic algorithm, semantic proximity.
Abstract: The actual problem of integrating IP is analysis, matching and mapping of their ontologies, which are constructed in different ways. The article provides a brief overview of the process of automatic construction of the initial versions of domain ontologies. An approach to the semantic integration of heterogeneous ontologies of complex information systems is proposed. The basic idea is to consider ontologies from different subject areas, and constructed in different ways. The article describes a method for calculating the semantic proximity of concepts, which allows one to quantify the similarity between concepts. The correspondences between the elements (concepts) of ontologies are divided into several components: lexical, attributive and relational. For each concept of one ontology, a set of relevant semantic concepts of another ontology is formed. For the purpose of ranking the elements of the resulting set, it is proposed to determine the threshold values of the proximity measure. A method for classifying the levels of proximity of concepts to establish their correct mapping is proposed. It should be noted that, excluding some error in comparing ontology concepts with the help of the software system, the main ontology core is adequately based on semantics and repeats the model constructed by experts. The high accuracy of the results is due to the use. The model of the integration process is applicable to a wide range of subject areas and is not difficult to implement.