【摘要】 针对编程教学中学习者思维固化与知识迁移困难的实际情况，分析目前大模型代码示范模式存在的认知断层问题，提出基于思维链的阶梯式引导策略，具体阐述如何通过特征词工程解构复杂任务为可操作的认知节点，构建具有时序逻辑的提示体系，在关键算法逻辑处设置反思性脚手架，提高代码注释密度与学习者调试自主性，形成词法—语法—语义—优化—程序实现的认知闭环，以数据科学与智能计算课程为例介绍教学实践并说明效果，为智能化编程教育提供可迁移的启发式教学范式。更多还原

【Abstract】 This study addresses the persistent challenges of cognitive rigidity and knowledge transfer barriers in programming education. By analyzing the cognitive disconnections inherent in current large-scale model code demonstration paradigms, a stepped guidance strategy based on Chain-of-Thought(CoT) reasoning is proposed. The framework systematically deconstructs complex programming tasks into operational cognitive nodes through feature keyword engineering, constructing a temporally-structured prompting framework that integrates sequential logic. Key algorithmic steps are augmented with reflective scaffolding to enhance code annotation density and learner-driven debugging autonomy. This approach forms a cognitive feedback loop spanning Lexical-Syntactic-Semantic-Optimization-Programing stages. Empirical validation is conducted through Data Science And Intelligent Computing courses, demonstrating the effectiveness of this heuristic pedagogical model in fostering transferable problem-solving skills. The study contributes a scalable, AI-assisted programming education paradigm grounded in cognitive science principles.更多还原