Design and Teaching Practice Application of LLM-RAG System

Abstract

This research designs and develops an intelligent question answering assistant system based on LLM-RAG technology. The system aims to provide students with a learning tutor supported by a self-built knowledge base. RAG technology can effectively address the hallucination problem in LLMs, while the self-built knowledge base function can overcome the limitation of insufficient real-time information in large models. Nine questions related to Python programming were designed to test the intelligent question answering assistant. After evaluation by three professional teachers in related fields, the system achieved good results in accuracy. Subsequent teaching practice in two classes showed that the improvement rate of class performance using the intelligent question answering assistant was significantly higher than that of the class without it, demonstrating its effectiveness in instructional support.