MODELING THE SEMANTIC INTERPRETATION OF PHRASEOLOGICAL UNITS IN AI-BASED TRANSLATION SYSTEMS: A LINGUOCULTURAL APPROACH

This study investigates the semantic interpretation of phraseological units within AI-based machine translation systems from a linguocultural perspective. Phraseological units are considered complex linguocultural entities that integrate semantic, metaphorical, cognitive, and cultural components. Although neural machine translation systems have achieved considerable success in recent years, they continue to face serious difficulties in interpreting idiomatic and culturally marked expressions. The research identifies the major limitations of AI translation systems in processing non-compositional meanings and culturally bound phraseological structures. To address these challenges, the study proposes a hybrid semantic interpretation framework that combines contextual embeddings, phraseological detection mechanisms, cultural annotation, and frame semantics. Comparative analysis of English and Uzbek phraseological units demonstrates that structural, semantic, and ethnocultural asymmetries significantly influence translation adequacy. The findings indicate that integrating linguocultural information into AI translation architectures substantially improves semantic accuracy and minimizes pragmatic distortion. The study contributes to the development of culturally adaptive and semantically sensitive translation models within the field of artificial intelligence and natural language processing.