از سیستم‌های آگاه به بافت تا مهندسی پرامپت هوش مصنوعی: چارچوبی مفهومی برای مدل‌سازی بافت در مطالعات ترجمه

Abstract

The emergence of Large Language Models (LLMs) has introduced a new paradigm in translation, making the quality of generated translations increasingly dependent on how contextual information is provided through prompts. Despite significant advances in AI-based translation, the selection and organization of contextual information still rely largely on users' individual experience and trial-and-error practices, while a coherent theoretical framework for the systematic modeling of context in interaction with these models remains absent. To address this gap, this study proposes a conceptual framework for context modeling in Translation Studies by drawing upon the principles of Context-Aware Systems and reinterpreting them within the domain of prompt engineering. This research is fundamental in nature and adopts a qualitative approach based on a comparative review and analysis of the literature in three domains: Translation Studies, Context-Aware Systems, and prompt engineering. Following the identification and comparison of contextual dimensions, a six-layer conceptual framework was developed, comprising linguistic context, discourse context, cultural context, audience context, translation task context, and interaction context with the language model. To demonstrate the practical applicability of the framework, a conceptual example of designing a context-aware prompt is presented, illustrating how contextual information can be systematically organized prior to interacting with a language model. The conceptual analysis suggests that the proposed framework provides an integrated structure for organizing contextual information and establishes a bridge between the theoretical foundations of Translation Studies, context modeling approaches in software engineering, and the practical application of prompt engineering. The proposed framework can serve as a foundation for developing standardized prompt design methodologies, documenting contextual information, and conducting future empirical research on AI-assisted translation.