Application of Statistic Methods for Development of Linguistic Support for Data Retrieval System

Problems of the data retrieval system development with natural language interface of requests are considered, among them, the preparation of dictionaries and search index taking into account syntactic structure of the document sentences. A method of the automatic creation of both the morphological and word-combination dictionary is suggested using statistical analysis of the sufficient amount of texts. The two-stage algorithm of the text syntax analysis is considered (using the simple formal and grammatical analysis at the first stage and the statistical refinement of the analysis results - at the second stage), and the text search algorithm as well, based on results of the two-stage algorithm application. Experimental estimations of the suggested methods operation quality are given.