Leshchynska I. Algebra-logical models of natural language structures and their application in artificial intelligence systems

The object of research is the process of parallel and automatic processing of natural language texts. The aim of research is to develop a complex of methods and models to formalize linguistic structures of natural language based on formal algebra of finite predicates and relational networks to extend the capabilities of intelligent user interface and subsequent realization of obtained models in intelligent natural language systems. The methods of research are logical analysis, algebra of finite predicates (for logical-mathematical constructs and creation of a full linguistic algebra apparatus), theory of linear logical operators (for development of a model of linear logic operators of the first and second kind), apparatus of multilayer predicates decomposition (for development of the relational networks theory), theory of comparing identification (for reasoning the structure of linguistic algebra), basic concepts of Boolean algebra and graph theory (for formalizing the trees of syntactic subordination). The equipment is a personal computer. Theoretical and practical result of research - actual scientific and practical problem of developing complex of methods and models to formalize linguistic structures of natural language based on formal algebra of finite predicates and relational networks for extending the capabilities of intelligent user interface was built. Scientific novelty - a model of linear logic operators of the first and second kind was offered first time, this operators describe the functioning of relational networks and based on the theory of linear logical operators that simplify the solution of logical equations system at each step of the relational network; a method for constructing logical structure of a relational associative network was developed first time, it is based on the use of hardware associative memory access, that allows easier access to data and increase the speed of finding solution; a method for constructing а predicate model of natural language sentence by introducing complementary and nonessential subject variables was developed first time. The method is based on the algebra of finite predicates and linguistic experiments and provide unambiguity of the sentence predicate model; linguistic algebra was gained further progress by the way of developing a method of constructing formulas of Boolean functions algebra that model the sentences of natural language, which simplifies the formalization of projective sentences. The results of the thesis, have been practically implemented in Design Bureau of Kharkov State Instrument Making plant of T.G. Shevchenko for support of design decisions during the development of the monitoring system of distribution electricity grid (MS of DEG) that is used for monitoring of main technical parameters of normal and emergency modes of electricity grid work 6 (10) kV, meteorological conditions (temperature and relative humidity) in the exploitation places, optimize search and localization of damaged places of overhead transmission lines; in PE "Energy" in order to develop computer complex for creation of common information environment for managing assets and resources of companies and to automate the process control technology services companies. Theoretical results of the thesis have been used in the educational process during preparation of lectures in such academic disciplines like "Theory of Intelligence" and "Bionics of Intelligence". Mathematical methods of formalizing of natural language structures, algorithms and a software system that developed in the thesis are designed for automated information systems with natural language intelligent interface and for for logical support of information structures design. Mathematical and program results can be used in systems of automatic text processing (exchange services, email).