Analysis of Associative and Sentimental Characteristics of Words for Revealing Hidden Emotions and Correlations

This paper investigates the relationship between text sentiment and associations, using data from open lexicographic resources: the Associative Dictionary and the Sentiment Dictionary of the Russian Language “KartaSlovSent”. The aim of the study is to explore how associative links can complement traditional sentiment analysis methods, uncover hidden emotional nuances, and refine the contextual interpretation of text. The paper proposes a data processing algorithm that includes stages of loading, merging, aggregating, and analyzing the distribution of sentiment associations. The results show that associative links can expand the sentiment lexicon, account for the context in which words are used, and identify local anomalies in emotional perception. A positive correlation was found between the numerical sentiment score of words and the share of positive associations. However, the study also revealed that the prevalence of neutral associations may reduce the accuracy of the analysis. Thus, the use of associations in sentiment analysis tasks is an effective auxiliary tool that can significantly improve the quality of emotional tone evaluation when integrated with traditional methods and machine learning models.