Evaluation of English-Russian Machine Translation of Technical Terms using Large Language Models

This paper investigates the application of Large Language Models (LLMs) to the translation of technical terms between English and Russian, addressing the challenge of accurately conveying domain-specific terminology across languages. Recognizing the limitations of existing machine translation models in dealing with specialized vocabulary, the study introduces a novel approach by constructing a dedicated corpus of technical terms and contexts. This corpus is compiled from a linguistic encyclopedic dictionary and enriched with contextual data sourced from Wikipedia, the Paracrawl parallel corpus, the TED Talks Corpus, and the CyberLeninka electronic library. The research methodologically explores various LLM prompting strategies when translating scientific texts for the English-Russian language pair. By conducting a comparative analysis of two research fields — linguistics and physics — the study intends to evaluate the impact of prompting on the quality of translations. The paper contributes to the field of computational linguistics by highlighting the effectiveness of integrating LLMs with domain-specific corpora and advanced prompting techniques for translating technical terminology.