基于近似子图的实时教学认知诊断模型设计与应用

Design and Application of the Cognitive Diagnosis Teaching Model Based on Approximate Sub-Graph

开展个性化教学是实现素质教育至关重要的措施和手段。实现个性化教学的主要手段是使用计算机辅助测验对学生知识结构进行诊断,然后根据诊断结果开展有针对性的资源推送和路径规划等补救教学活动,其核心问题是教学认知诊断模型的构建。当前常用的基于RSM和AHM的教学认知诊断模型虽然能够对学生的知识状态进行高效、准确的诊断,但其生成规则空间的代价较高,在小规模实时诊断应用中效率较低。近似子图是领域知识图在测试项目下的相关子图,其顶点集只包含测试项目中涉及到的认知属性,其边集包括顶点集在领域知识图中直接依赖关系的投影,以及顶点集在领域知识图中间接依赖关系的模拟。近似子图可以降低规则空间规模的量级,满足其构造的实时性要求。基于近似子图的教学认知诊断模型在"Java语言程序设计"课程中的教学实验证明:该模型可以降低构建RSM规则空间的时间代价,在保证诊断准确率的前提下达到课堂实时认知诊断的目的,及时帮助学生发现自身知识缺陷并开展有针对性的补救学习,最终有效改善学生学习效果。针对学生的调查问卷结果也显示,该教学模式较传统课堂更为轻松和高效。

Personalized teaching is the crucial measure and means of realizing quality education. The main method to realize personalized teaching is computer-aided testing, which is used to diagnose students＇ knowledge structure, and the diagnosis results provide the basis of carrying out targeted remedial activities, such as resourcepush and path planning. The core problem of knowledge diagnosis is to construct a cognitive diagnosis teaching model. Although commonly used cognitive diagnosis teaching models based on RSM and AHM can diagnose students＇ knowledge state efficiently and accurately, the cost of generating rule space is high, and their efficiency is low when applying them into small-scale real-time diagnosis. Approximate sub-graph is the correlation sub-graph of domain knowledge graph under test items, the vertex set of which only contains cognitive attributes involved in test items, and the edge set includes the projection of direct dependency of vertex set and the simulation of indirect dependency of vertex set on domain knowledge graph. Approximate sub-graph can reduce the scale of rule space,which meets the real-time requirement of rule space construction. Approximate sub-graph based cognitive diagnosis teaching model was applied in Java programming course, and the experimental results show that this model reduces the time cost of constructing RSM rule space, makes small-scale real-time cognitive diagnosis works in classroom without any loss of diagnosis accuracy, and makes students＇ knowledge defects more clear which helps students perform more targeted remedial learning, thus to improve the learning effect of students eventually.According to the results of the questionnaire, students participating in the experiment generally believe that this teaching mode is more relaxed and efficient than traditional mode.