基于DeepSeek与RAG技术的教学智能问答系统研究

打开文本图片集

中图分类号：TP311.1 文献标识码：A 文章编号：2096-4706（2025）24-0045-06

Abstract：Intellgent question-answering systems based on Large Language Models （LLMs）can provide interactive leamingsupport in software professional teaching，buttheyhave problems including halucination generation，knowledge lag， and insuffcient professonal depth. Based onRetrieval-Augmented Generation （RAG）technology，this paper constructs an intellgent question-answering system suitableforsofware professonal teaching bycombining DeepSeek modeland multimodal data procesing methods.Byintroducing an extemal knowledge base tooptimize the“retrieval-generation”processandusing quantitative indicators such as recall （85%） and precision （83%） to evaluate the system performance，this paper aims to reduce theincidenceofhallucinations，improve knowledgetimelinessandprofessonaldepth，andprovideaneficientandaccurate intelligent auxiliary tool for software professional teaching.

Keywords：Large Language Model; Retrieval-Augmented Generation; teaching intelligent question-answering system

0 引言

软件专业教学中，学生对编程、算法等知识的个性化需求日益增长，传统教学模式已难以满足全时段的交互指导需求。（剩余8796字）