“TILSE” Framework for RAG-Based AIGC Feedback Prompts: A Modular and Personalized Intelligent Feedback Generation Method
Beibei Peng, Xindi Wang, Lei Xu · 2024
This paper introduces the TILSE grading prompt framework, integrating RAG (Retrieval-Augmented Generation) technology to address the challenge of generating accurate and personalized feedback in educational settings. The TILSE framework's modular design, comprising Task, Input, Logic, Style, and Example modules, allows for flexible and contextually relevant prompt generation. By dynamically retrieving pertinent knowledge, RAG technology enhances the precision and adaptability of feedback, making it more tailored to individual student needs. Experiments with ChatGPT 4.0 demonstrate that the TILSE framework significantly outperforms traditional methods, particularly in complex educational scenarios, by providing more accurate and personalized feedback. This research offers a novel solution to the limitations of existing feedback systems and contributes to the advancement of AI-driven educational tools.