Sentiment Analysis of Social Networks Messages. Tretyakov, E., Savić, D., Korpusenko, A., Ionkina, K. (2022). In: Klimov, V.V., Kelley, D.J. (eds) Biologically Inspired Cognitive Architectures 2021. BICA 2021. Studies in Computational Intelligence, vol 1032. Springer, Cham.
In the modern era of artificial intelligence and machine learning, data mining is becoming an important tool for determining public opinion and social research. In this regard, sentiment analysis is a new method of studying public opinion, in particular, as a nontrivial approach to the analysis of political texts. This paper examines the nature of sentiment analysis in political texts, identifies the problems which researchers face when analyzing political texts, and identifies the difficulties that affect the accuracy of results. The aim of this study is to determine the relevance of sentiment analysis in the analysis of political texts. It presents an ongoing work that is developing an algorithm combining a lexical-oriented approach with machine learning, that studies stylistic devices (e.g., sarcasm, irony and hyperbole), and provides options for determining the sentiment of texts in sentences containing these stylistic devices. As a result of the experiments, patterns that affect the accuracy of the analysis result are identified and ways to handle them are suggested in order to improve the accuracy of the results. Options for determining the sentiment of texts in sentences containing stylistic devices are provided as a contribution to the scientific field.