数据驱动的数据结构精准诊断与个性化学习实践研究——基于PTA与学习通的混合智能教学探索

Data-Driven Precision Diagnosis and Personalized Learning Practice for Data Structures—Hybrid Intelligent Teaching Exploration Based on PTA and Xuexitong

摘要: 针对《数据结构》课程教学中长期存在的“指针恐惧症”、“递归理解难”、“调试效率低”、“分层教学缺失”四大核心痛点，本研究构建了“PTA数据感知–混合智能诊断–学习通平台干预”三位一体的智能化教学体系。通过PTA平台自动化采集学生学习行为数据，运用聚类分析与开源大语言模型(如ChatGLM-6B)进行学习困境诊断，依托学习通平台实现精准化教学干预。在一项为期16周的准实验研究中，实验组(60人)采用上述模式，对照组(60人)接受传统教学。结果显示：实验组调试时间缩短33.3%，指针与递归题型正确率从55.3%和49.7%分别提升至76.8%和71.2%，教师重复答疑工作量减少41.7%。大语言模型在153个代码错误样本上的诊断准确率达86.9%，K-Means聚类将学生划分的三个学习群体特征可解释性良好。半结构化访谈显示，实验组学生普遍认可诊断推送的针对性和实用价值。本研究为计算机类专业课程的教育数字化转型提供了可复制的实践案例。

Abstract: To address the four long-standing core pain points in the teaching of the Data Structures course—namely “pointer phobia,” “difficulty in understanding recursion,” “low debugging efficiency,” and “lack of tiered instruction”—this study constructs a three in one intelligent teaching system that integrates “PTA data perception, hybrid intelligent diagnosis, and Xuexitong platform intervention.” By automatically collecting students’ learning behavior data through the PTA platform, employing clustering analysis and an open source large language model (e.g., ChatGLM 6B) to diagnose learning difficulties, and relying on the Xuexitong platform for precise instructional intervention, a 16 week quasi experimental study was conducted with an experimental group (60 students) adopting the above model and a control group (60 students) receiving traditional teaching. The results show that the experimental group reduced debugging time by 33.3%, increased accuracy on pointer and recursion related problem types from 55.3% and 49.7% to 76.8% and 71.2%, respectively, and cut teachers’ repetitive question answering workload by 41.7%. The large language model achieved a diagnostic accuracy of 86.9% on 153 code error samples, and the K Means clustering produced three learning groups with good interpretability. Semi structured interviews indicated that students in the experimental group generally recognized the relevance and practical value of the diagnostic feedback and resource recommendations. This study provides a replicable practice case for the digital transformation of education in computer related programs.