Interpreted sentiment analysis for the Ukrainian language based on rules

Abstract

Sentiment analysis plays a vital role in natural language processing (NLP), enabling automated evaluation of emotional tone in text across multiple languages. Yet, popular tools like VADER often fall short when applied to languages with complex morphology and syntax, such as Ukrainian. This study presents an enhanced rule-based sentiment analysis algorithm tailored specifically for Ukrainian-language content, addressing the limitations of generic, English-centric models. The algorithm incorporates an expanded lexicon that includes the EMOLEX sentiment dictionary, polarity scores, emoji sentiment mappings, and intensity modifiers to improve sentiment detection. It also leverages advanced techniques such as dependency parsing and position-aware scoring to better capture contextual nuances. These enhancements are necessary for correctly deciphering Ukrainian's distinctive linguistic structures, which frequently present difficulties for conventional sentiment analysis systems. The algorithm was evaluated using datasets in the Ukrainian language and compared to VADER. The custom model performs noticeably better than VADER, according to the results, especially when it comes to detecting strongly positive or negative sentiments. Because they provide greater accuracy and contextual awareness, these results emphasise the value of language-specific tools for non-English content. Even though the results are encouraging, more work is necessary. In order to create a hybrid system that can handle increasingly complex and ambiguous expressions with even higher accuracy, future research may investigate integrating AI techniques, such as machine learning and deep learning.