ScholarWorks@경상국립대학교

초록

This study aims to demonstrate the linguistic significance of classifying Russian emotional adjectives based on word embeddings. To achieve this, the study examines the types of emotional vocabulary. Secondly, it reviews previous research on the semantics of Russian emotional vocabulary. Thirdly, it explores semantic associates based on word embeddings. Fourthly, the study groups Russian emotional adjectives using word embedding-based semantic associates and analyzes differences between these results and those of traditional lexical semantic research. Finally, it discusses the potential role of word embeddings in the semantic analysis of emotional vocabulary. This study utilized three models built by the Language Technology Group at the University of Oslo. To evaluate the performance of the models, an analogy evaluation was conducted to assess how well the language models understand the semantic relationships between words. As a result, it was confirmed that a sufficient quantity and quality of data, accurate preprocessing, and effective parameter combinations for learning are necessary to construct high-quality models. Through the analysis of the similarity between "тоскливый" and "грустный", it was examined how the synonyms of these adjectives are represented. Therefore, exploring the semantic relationships of emotional vocabulary through word embeddings can serve as a valuable tool in linguistic research, demonstrating the complementary relationship between linguistic intuition and machine learning results.

키워드