STUDY ON SENTIMENT ANALYSIS METHODS FOR TEXT DATA

Abstract

The relevance of the research topic is determined by the avalanche-like growth of unstructured text data on the Internet and the need for effective methods of tonality analysis. The purpose of the work is to systematically study the current state of the tonality analysis methodology, compare the leading approaches and outline further prospects. The article analyzes in detail popular Python libraries for natural language processing – NLTK, spaCy, TextBlob, Gensim. The comparison was made according to the criteria of computational efficiency, ease of use, flexibility of feature extraction, and customization options. The methodological core of the study is an experimental comparison of NLTK and TextBlob for tonality classification of Ukrainian-language texts. Estimates may vary depending on specific usage scenario and settings. NLTK, where it can be more accurate when configured correctly, but requires more effort to configure. TextBlob, on the other hand, is easier to use, but may be less accurate for specialized tasks. The results proved the superiority of TextBlob in speed and NLTK in accuracy. Tonality analysis has a huge potential for improving analytical capabilities in many areas – from optimizing business processes to countering the spread of fake news. Further research should focus on the development of specialized solutions for specific applied tasks. Prospects for improving the ethical principles of text analysis, taking into account the linguistic and cultural context, as well as the integration of the tonality analysis functionality into decision support systems have been determined.