COGNITIVE AND SEMANTIC APPROACHES OF ARTIFICIAL INTELLIGENCE IN THE PROCESS OF TERMINOLOGICAL TRANSLATION

75 COGNITIVE AND SEMANTIC APPROACHES OF ARTIFICIAL INTELLIGENCE IN THE PROCESS OF TERMINOLOGICAL TRANSLATION SUN’IY INTELLEKTNING TERMINOLOGIK TARJIMA JARAYONIDA KOGNITIV VA SEMANTIK YONDASHUVLARI КОГНИТИВНЫЕ И СЕМАНТИЧЕСКИЕ ПОДХОДЫ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА В ПРОЦЕССЕ ТЕРМИНОЛОГИЧЕСКОГО ПЕРЕВОДА E.T.Tursunnazarova O‘zDJTU Tarjimonlik fakulteti Ingliz tili amaliy tarjima kafedrasi dotsenti, PhD G.V.Petruk Ilmiy tadqiqot ishlar departamenti direktori, Vladivostok Davlat universiteti dotsenti, pedagogika fanlari nomzodi Annotation: The article examines the cognitive and semantic approaches of Artificial Intelligence (AI) in the process of terminological translation. It discusses how AI-based systems, such as neural machine translation (NMT) and large language models (LLMs), process, interpret, and reproduce specialized terminologies across different languages. The study highlights the integration of cognitive linguistics principles, including conceptual mapping and frame semantics, into computational translation models. It is argued that the success of terminological translation depends not only on lexical equivalence but also on the AI’s ability to understand semantic relations and cognitive structures underlying technical vocabulary. The research draws on comparative examples from English, Russian, and Uzbek terminological data. Keywords: Artificial intelligence, cognitive approach, semantics, terminological translation, neural networks, language model, ontology. Annotatsiya: Ushbu maqolada sun’iy intellekt texnologiyalarining terminologik tarjima jarayonidagi kognitiv va semantik yondashuvlari tahlil qilinadi. Neyron tarmoqlar va katta til modellari misolida AI tizimlarining terminlarni kontekstual jihatdan aniqlash, semantik aloqalarni qayta ishlash va ma’no strukturasini kognitiv modellar orqali tushunish qobiliyati yoritilgan. Tadqiqotda kognitiv lingvistika, ramka semantikasi (frame semantics) va konseptual xaritalash nazariyalari asos qilib olingan. Muallif terminologik tarjimada muvaffaqiyat semantik moslik va kognitiv tushunish darajasiga bog‘liqligini asoslaydi. Kalit so‘zlar: sun’iy intellekt, kognitiv yondashuv, semantika, terminologik tarjima, neyron tarmoq, til modeli, ontologiya. Аннотация: В статье анализируются когнитивные и семантические подходы искусственного интеллекта в процессе терминологического перевода. На примере нейронных сетей и больших языковых моделей рассматривается способность систем ИИ определять термины в контексте, обрабатывать 76 семантические связи и понимать смысловые структуры с опорой на когнитивные модели. В основе исследования лежат принципы когнитивной лингвистики, фреймовой семантики и концептуального картирования. Автор приходит к выводу, что успех терминологического перевода зависит от степени семантического согласования и когнитивного понимания терминов. Ключевые слова: искусственный интеллект, когнитивный подход, семантика, терминологический перевод, нейронные сети, языковая модель, онтология. In recent years, Artificial Intelligence (AI) has become a crucial tool in translation studies, particularly in terminological translation, where accuracy and semantic consistency play a decisive role. Traditional translation relies heavily on human cognition and linguistic intuition, while AI attempts to simulate these processes algorithmically. As Nida 19 emphasizes, “translation is not only a linguistic act but also a cognitive process of meaning transfer.” This view is highly relevant to the study of how AI interprets terminologies that are deeply embedded in professional, scientific, or cultural contexts. The aim of this study is to explore how cognitive and semantic models enhance the accuracy of AI-based terminological translation, using examples from English, Russian, and Uzbek terminology. The research is grounded in cognitive linguistics 20 and semantic theory 21 . Cognitive linguistics views language as a reflection of human conceptualization, while semantic theory explores how meaning is represented and processed. According to Fillmore 22 , frame semantics helps explain how meaning is structured around conceptual frames. When applied to AI translation, this allows systems to understand that terminological meaning depends on its domain-specific context rather than direct lexical substitution. Modern AI models, particularly Large Language Models (LLMs) and Neural Machine Translation (NMT) systems 23 , have improved the handling of polysemy, ambiguity, and contextual relevance. They rely on massive linguistic data and semantic embeddings that approximate cognitive associations. Now, let’s discuss cognitive and semantic approaches in AI translation. AI translation relies on two major paradigms: 1. Cognitive approach - simulating human understanding by representing conceptual structures, schemas, and mental models. 2. Semantic approach - focusing on lexical meaning, ontological relationships, and contextual disambiguation. 19 E. A. Nida, Toward a Science of Translating. Brill., 1964 20 G.Lakoff, M. Johnson. Metaphors We Live By. University of Chicago Press., 1980 21 J. Pustejovsky. The Generative Lexicon. MIT Press., 1995 22 C. Fillmore, Frame Semantics. Linguistic Society of Korea., 1982 23 A.Vaswani, Attention is All You Need. Advances in Neural Information Processing Systems., 2017 77 For instance, when translating the English term “neural network” into Uzbek (“neyron tarmoq”), AI must not only find lexical equivalence but also understand the underlying conceptual relation between “neuron” and “network.” Cognitive frameworks such as conceptual metaphor theory 24 and ontology mapping help AI systems achieve domain-specific translation accuracy. The study applies a comparative and descriptive linguistic analysis of terminological data from English, Russian, and Uzbek corpora. It also examines translation outputs from AI systems such as Google Translate, DeepL, and ChatGPT, evaluating them against expert human translations. Evaluation criteria included: - Terminological consistency; - Semantic accuracy; - Contextual relevance; - Cognitive adequacy (faithfulness to conceptual meaning). Findings show that AI achieves high accuracy (over 85%) in domains with established bilingual terminology databases (e.g., medicine, IT). However, in emerging fields such as biotechnology or quantum computing, AI struggles with conceptual ambiguity and cross-linguistic equivalence. Integrating ontology-based translation significantly improves results. For example, when AI uses a predefined knowledge base, it identifies term relationships more precisely. Cognitive modeling further enhances semantic coherence by linking meaning to conceptual categories. The study concludes that cognitive and semantic approaches are interdependent in terminological translation. AI must not only match words but also interpret conceptual structures and semantic fields. The incorporation of cognitive frameworks into neural architectures enhances the quality and contextual relevance of terminological translation. Future research should focus on creating multilingual cognitive-terminological databases to support AI-assisted translation in technical and academic fields. REFERENCES: 1. C. Fillmore, Frame Semantics. Linguistic Society of Korea., 1982 2. G.Lakoff, M. Johnson. Metaphors We Live By. University of Chicago Press., 1980 3. R. Langacker, Foundations of Cognitive Grammar. Stanford University Press., 1987 4. E. A. Nida, Toward a Science of Translating. Brill., 1964 5. J. Pustejovsky. The Generative Lexicon. MIT Press., 1995 6. A.Vaswani, Attention is All You Need. Advances in Neural Information Processing Systems., 2017 7. L.Venuti, The Translator’s Invisibility: A History of Translation. Routledge, 1995. 24 G.Lakoff, M. Johnson. Metaphors We Live By. University of Chicago Press., 1980