多智能体协同赋能的实践课程教学模式构建——以数据分析类实践课程为例

Construction of a Multi-Agent Collaboration-Empowered Teaching Approach for Practice-Oriented Courses—A Case Study of Data Analytics Practice-Oriented Courses

摘要: 针对数据分析类实践课程普遍存在的教学内容更新滞后、真实业务场景案例不足、个性化指导缺失以及过程性评价难以量化等问题，研究以数据挖掘实践课程为例，通过多个智能体协同，构建覆盖课前、课中、课后的全链路智慧教学模式。依托超星学习通AI工作台，集成教学设计专家、伴学助手、学情分析、AI实践与智能评测等多个智能体，重构实践课程教学流程，形成“情境创设–原理探究–AI思辨–实践育人–智能评价”的递进式教学体系。实施层面，通过智能体生成贴近产业场景的真实数据集缓解案例获取困难问题；依托7 × 24小时代码纠错与实时反馈机制提供个性化指导；通过智能体自动批阅与教师复核实现多维度过程性评价。实践结果表明，该模式显著提升教学精准度与学生学习投入度，学生项目平均成绩达92.47分，优秀率提升至34.02%。整体上，该模式推动课程教学由“经验驱动”向“数据与智能体协同驱动”转型，为数据分析类实践课程的数字化改革提供了可复制的实践路径。

Abstract: Addressing common issues in data analytics practice courses, such as outdated teaching content, insufficient real-world business cases, lack of personalized guidance, and difficulty in quantifying process-based assessment, this study takes a data mining practice course as an example and constructs a full-process smart teaching mode covering pre-class, in-class, and post-class stages through the collaboration of multiple intelligent agents. Based on the AI Workspace of the Chaoxing Learning Platform, multiple intelligent agents, including a teaching design expert agent, a learning companion agent, a learning analytics agent, an AI practice agent, and an intelligent assessment agent, are integrated to reconstruct the teaching process of practice-oriented courses, forming a progressive teaching framework of “scenario creation - theoretical exploration - AI-based critical thinking - practice-based education-intelligent evaluation.” In implementation, intelligent agents generate realistic datasets closely aligned with industry scenarios to alleviate difficulties in obtaining suitable cases. A 24/7 code debugging and real-time feedback mechanism provides personalized guidance. Automated grading by intelligent agents combined with teacher review enables multidimensional process-based assessment. Practical results show that the proposed mode significantly improves teaching precision and student engagement. The average project score reached 92.47 points, and the excellence rate increased to 34.02%. Overall, the model shifts course teaching from being experience-driven to being driven by collaboration between data and intelligent agents, offering a replicable approach to the digital transformation of data analytics practice courses.