一句话：提出基于 RAG 的虚拟助教，从课程教材中提取知识，为 CS1 问题提供可验证引用。
CS1 学生常难以在分散资源中找到适合课程语境的答案，也可能因隐藏的知识缺口而不知如何提问。本文提出一种检索增强生成（RAG）虚拟助教（VTA），从课程材料中提取知识，用与 CS1 对齐的解释回答考试题和概念题，并给出可验证的教科书引用。实验表明，该助教能提供可靠、完整的回答，并正确链接到相关教科书章节。使用日志与调查显示，学生更多提出概念性问题而非问题求解类问题，89% 的学生认为该助教有帮助，表明基于教材的 LLM 支持能促进学生对核心概念的深入参与。
Students in CS1 often struggle to find course-appropriate answers across scattered resources and may not know how to ask questions due to hidden knowledge gaps. We present a Retrieval-Augmented Generation (RAG) Virtual Teaching Assistant(VTA) that extracts knowledge from course materials to answer exam and conceptual questions with CS1-aligned explanations and verifiable textbook citations. Experiments show reliable, complete responses with correct links to relevant textbook sections. Usage logs and surveys indicate that students ask more conceptual than problem-solving questions, and 89% report that the assistant is helpful—suggesting that grounded LLM support can encourage deeper engagement with core concepts.