Kamenieva I. The models of relational networks of brain-like structures and their application in artificial intelligence

Object of research - the process of formal description of human intellectual activity using the models of relational logic networks of brain-like structures. Subject of research - The models of relational networks of brain-like structures and their application in artificial intelligence. Theoretical and practical results of research are to develop formal inflection model of short adjectives, formal generalized model of Russian inflection of adjectives and the method of separation of the finite predicate. Their implementation in practice enhances the efficiency of processing natural language information and the significant strengthening of intellectual capacities of human; the practical value of the work is to develop software and hardware and software infrastructure, an experimental study that confirms the effectiveness of the proposed approach. The aim is to improve the efficiency of information processing in artificial intelligence systems by developing a formal inflection model of short adjectives, the generalized model of Russian inflection of adjectives and the further development of the bundle method for finite predicate of brain-like structures. It is modeling based on the algebra of finite predicates by relational logical networks in artificial intelligence systems. They are focused on parallel processing of information and subsequent software implementation. The scientific novelty - formal inflection model of short adjectives was first proposed. It is characterized by introducing a new subject variables that are described by systems of binary relations and set by formulas of corresponding predicates. The class of problems which are solved by a relational logical network was extended by proposed model. First proposed and studied formal generalized model of Russian inflection of adjectives, which provides complete conjugation of non possessive adjectives, potential word forms of full non possessive adjectives that are presented by sound-form and makes the inflection of short form of adjectives in Russian. The model allows to solve problems of analysis, synthesis and normalization of natural language information and improve the quality of search in the natural-language systems. The method of separation of the finite predicate for automated model building sound declination form full potential not possessive Russian adjectives as a relational logic network was further developed. It is used to solve morphological problems of natural language information processing. The results of the dissertation are introduced to SE "Sytenko Institute of Spine and Joint Pathology, AMS Ukraine", to firm LLC "Staleks" and to the learning process of the Kharkov National University of Radio Electronics. The scientific and practical results of the dissertation can be used in organizations that deal with the problems of development, research, implementation and operation of intelligent systems of processing the grammatical structure of the text, identification and modeling systems of human speech.