نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Conversational question and answer systems have evolved significantly with the advent of large language models. However, these advances have mostly benefited high-resource languages such as English, while for low-resource languages such as Persian limited resources have been developed. This paper introduces a novel framework augmented with large language models that is specifically designed to fill this linguistic gap. The proposed architecture has six components: “input processing” for language-specific management, “adaptive large language model core”, “knowledge enhancement” for cross-lingual mapping, “context management” for efficient conversational navigation, “response generation” with cultural adaptation, and “human feedback” for continuous improvement. Unlike existing approaches that treat low-resource languages such as Persian as a secondary consideration, the proposed framework incorporates cultural and linguistic considerations throughout the process. The results of the evaluation of the proposed framework by experts are also presented to determine how this framework can address the challenges faced in low-resource languages, including limited training data, morphological complexity, cultural subtleties, and computational limitations.
کلیدواژهها English
