Abstract:
This paper introduces an approach to building decision support systems based on a semiotic domain model and natural language processing methods. The knowledge base of this model is a text corpus of linguistic information obtained from the Internet. The text corpus is relevant to the subject domain in which the subjective semiotic model of the situation is constructed. A method for solving the inverse problem in a semiotic system is proposed. The obtained solutions are interpreted in the subject domain using a semantic calculator. The semantic calculator extracts generic relations from the text corpus based on lexico-syntactic patterns and determines the frequency of joint occurrence of words in the solution based on the distributive analysis of the text corpus. The generalized structures of monitoring and decision-making sub-systems with the semiotic model of the situation and natural language processing methods are described. A software layout of the decision-making subsystem is developed. The effectiveness of this approach is demonstrated by experiments.
Keywords:decision-making, semiotic system, subjective model, natural language processing, distributive analysis.
Fulltext:
PDF file (1881 kB)