REVIEW OF METHODS FOR DETERMINING THE TONATION OF TEXTS IN NATURAL LANGUAGES

Abstract

The analysis of sentiment in user comments finds application in many areas, such as evaluating the quality of goods and services, analyzing emotions in messages, and detecting phishing advertisements. There are numerous methods for analyzing the sentiment of textual data in the Russian language, but automatic sentiment analysis of Russian-language texts is much less developed than for other major world languages. This article is part of a broader study on the creation of an information system for detecting dangerous content in the cyberspace of Kazakhstan. The purpose of this article is to provide an analytical review of the different approaches to sentiment analysis of Russian-language texts and to compare modern methods for solving the problem of text classification. Additionally, the article seeks to identify development trends in this area and select the best algorithms for use in further research. The review covers different methods for text data preprocessing, vectorization, and machine classification for sentiment analysis of texts, and it concludes with an analysis of existing databases on this topic. The article identifies some of the main unresolved problems in sentiment analysis of Russianlanguage texts and discusses planned further research.