生成式人工智能背景下软件框架课程教学评价重构——基于2021~2025年课程作业的诊断与方案设计

Reconstruction of Teaching Assessment in Software Framework Courses in the Context of Generative Artificial Intelligence—Diagnosis and Reform Design Based on Course Assignments from 2021 to 2025

摘要: 生成式人工智能能够参与代码生成、测试和文档编写，使软件框架课程中以项目功能和运行结果为主的评价方式难以充分反映学生的真实学习情况。本文以《软件工具与环境(高级)》课程2021~2025年学生提交材料为基础，分析课程任务演进及现有评价中存在的问题。结果表明，课程任务逐步由基础框架练习扩展到综合项目，但现有评价仍偏重最终成果，对学生实际理解框架运行机制过程关注不足。本文以机制理解为核心，重构课程任务和评价方式，通过递进式实践任务和多源学习证据考察学生的机制解释、架构判断、代码审查和任务迁移能力，为生成式人工智能环境下软件框架课程教学改革提供参考。

Abstract: Generative artificial intelligence can participate in code generation, testing, and documentation, making it difficult for evaluation methods that mainly rely on project functionality and execution results to fully reflect students’ actual learning in software framework courses. Using student submissions from the course Software Tools and Environments (Advanced) between 2021 and 2025, this study analyzes the evolution of course tasks and the limitations of existing assessment practices. The results show that course tasks have gradually expanded from basic framework exercises to comprehensive projects, while current assessment still places greater emphasis on final products and provides limited evidence of students’ understanding of framework runtime mechanisms. To address this issue, the study takes mechanism understanding as the core learning objective and redesigns course tasks and assessment methods. Through progressive practical tasks and multiple sources of learning evidence, students’ abilities in mechanism explanation, architectural judgment, code review, and task transfer are assessed. The proposed approach provides a reference for the reform of software framework courses in the context of generative artificial intelligence.