Підхід до розробки інформаційної системи автоматизації перекладу на основі нейромережевих моделей

Abstract

The paper considers approaches to building translation automation information systems based on neural models. The functional capabilities of CAT tools, translation management platforms, machine translation systems, and localisation tools are analysed. It is determined that existing solutions usually do not provide full integration of neural translation, translation memory, terminology bases, and quality assurance tools within a single environment. An approach to developing an information system that combines NMT/LLM translation, post-editing, QA verification, project management, and user role management modules is proposed. The expected result is increased translator productivity, terminology consistency, and final translation quality.