Conversational speech analysis based on the formalized representation of the mental lexicon

Abstract

This article is dedicated to the study of the possibilities of formalization and modeling with the help ofmodern approaches in the field of artificial intelligence such a structure of human consciousness as “the mental lexicon”. According to experts in the field of psycholinguistics, it is a large graph of concepts and lexemes of a language, related to each other by semantic connections, which plays an essential role in the processes of generation and perception of speech by a human. The usage of semantic networks for modeling this structure allows using the advantages of this form of representation in full for solving problems of natural speech understanding by technical systems. This approach also implies the need to develop methods of analysis and processing of the speech signal, which precede the stages of semantic processing, closer integration of these stages and, in a sense, the erasing of the boundaries between them, taking into account how this occurs in the process of speech perception by the human brain. This article is concerned with the possibilities to show the advantages of such a complex approach to the analysis of speech messages through its implementation based on the Open Semantic Technology for Intelligent Systems (OSTIS).