Вероятностно-статистические модели в производстве автороведческой экспертизы русскоязычных текстов

Abstract

The article presents the experience of developing a computerized text authorization model and its adaptation to the needs of identification and diagnostic authoring expertise of Russian texts. The purpose of the study is to demonstrate the possibilities of identification authoring expert examination of texts through automatic text processing based on the integrated application of probabilistic-statistical methods. The article describes one more stage of testing an improved version of the CAT program (computerized text authorization) - an experiment to determine the relative frequencies of the ratio of certain linguistic elements (calculation of correlation coefficients) in several samples of compared texts using a set of different-level parameters - B. N. Golovin's coefficients, supplemented by the text clarity indexes existing in scientific prac-tice and applied field ("Flesch-Kincaid Index", "FOG-Index") and a number of other parameters. The research materials consist of primary databases of texts of Russian classics (works by L. N. Tolstoy, N. V. Gogol, I. S. Turgenev.

В статье рассматриваются возможности идентификационного автороведческого экспертного исследования текстов посредством автоматической обработки на основе комплексного применения вероятностно-статистических методов. Материалом исследования являются произведения Л. Н. Толстого, Н. В. Гоголя, И. С. Тургенева.