智能体驱动的机器学习课程“教–学–评–改”教学改革与实践

The “Teaching - Learning - Assessment - Revision” Teaching Reform and Practice of the Agent-Driven Machine Learning Course

摘要: 针对机器学习课程教学中存在的教学资源更新滞后、知识问答缺乏精准性、个性化学习支持不足、教学评价反馈滞后等现实问题，本文提出基于智能体技术的“教–学–评–改”四位一体教学改革方案。通过在Coze平台构建四类教学智能体，教学资源建设智能体实现“搜索–去重–分类–摘要–推荐–日报”全流程自动化资源采集与课程映射，RAG课程学习智能体实现基于课程知识库的精准问答与学习辅导，个性化学习智能体实现“诊断–推荐–训练–反馈”闭环因材施教，教学评价与持续改进智能体实现学情自动统计与教学分析报告生成，形成覆盖教学全流程的智能支持体系。本文详细阐述了各智能体的设计思路、技术架构与实现流程，为人工智能赋能高校课程教学改革提供了可复制、可推广的实践路径。

Abstract: To address the practical issues in machine learning course teaching, such as lagging updates of teaching resources, lack of precision in knowledge Q&A, insufficient personalized learning support, and delayed teaching evaluation feedback, this paper proposes a “teaching - learning - evaluation - correction” four-in-one teaching reform scheme based on agent technology. By constructing four types of teaching agents on the Coze platform, the teaching resource construction agent realizes the full-process automation of resource collection and course mapping, including “search - deduplication - classification - summary - recommendation - daily report”. The RAG course learning agent achieves precise Q&A and learning guidance based on the course knowledge base. The personalized learning agent implements a closed-loop “diagnosis - recommendation - training - feedback” for individualized teaching. The teaching evaluation and continuous improvement agent facilitates automatic student learning statistics and the generation of teaching analysis reports, forming an intelligent support system covering the entire teaching process. This paper elaborates on the design ideas, technical architecture, and implementation processes of each agent, providing a replicable and scalable practical path for AI-enabled curriculum teaching reform in universities.