A ontological approach to determining the subject matter of scientific text

The paper proposes an approach to determination the subject matter of scientific text, using the domain ontology. It presents the original principle of building applied sciences ontologies, in which the ontology tree contains three required branches: the "Fundamental theory", "Applied theory", "Applications". Topics are classes of the ontology and terms of the matching topics are instances of classes. We describe the structure of the control theory ontology and present fragments of this ontology and its dictionary. It is assumed that the scientific text relevant to the topic, if it contains terms of this topic. A method is proposed for evaluating of the degree of scientific text relevance to different topics based on counting the number of the terms occurrences of these topics in the document. The result of this method is the “the document profile” - vector of document relevance to ontology topics. We describe the automatic system for analysis the subjects of scientific texts from the field of the control theory and practice developed on the basis of the proposed approach. The linguistic problems of the terms search are considered. We present some statistics after processing of thematic sections of the journal "Automation and Remote Control" and an example of building a profile for selected article. Possible directions to improve the estimates of relevance are noted.