MODELING THE SEMANTIC INTERPRETATION OF PHRASEOLOGICAL UNITS IN AI-BASED MACHINE TRANSLATION SYSTEMS: A COGNITIVE AND HYBRID APPROACH

Abstract

 This article provides a comprehensive analysis of the problem of semantic interpretation of phraseological units (idiomatic expressions) in artificial intelligence (AI)-based machine translation systems, with a particular focus on Uzbek-English language pairs. Phraseological units pose significant challenges due to their morphological complexity, polysemy, and cultural specificity.

The study compares different translation approaches, including rule-based, statistical phrase-based, neural machine translation, and large language models, highlighting their strengths and limitations. It is shown that these systems often struggle with capturing contextual meaning and tend to produce literal translations.

The research emphasizes the importance of integrating cognitive linguistics and conceptual metaphor theory into translation models. A set of six English-Uzbek phraseological units is analyzed contextually to demonstrate translation strategies.

Finally, the paper proposes solutions such as preprocessing, paraphrasing, phraseological dictionaries, and hybrid models to improve translation quality. The findings suggest that combining multiple approaches can significantly enhance the semantic accuracy of idiomatic translation. Future research directions are also outlined.