Abstract: Aiming at the low pertinence and efficiency problem caused by massive exercises, a personalized exercises Recommendation algorithm based on Knowledge Hierarchical Graph（ReKHG） is proposed. Firstly, in consideration of the characteristic of system architecture, a weight map is built, which characterizes the knowledge hierarchical relations. This map can reflect the hierarchical relationship among knowledges efficiently. Then, a personalized exercise recommendation algorithm is proposed according to students’ mastery of knowledges and their hierarchical graph. The ReKHG algorithm acquires students’ weakness of knowledge mastery by updating knowledge losing-rate matrix of students. Experimental results show that ReKHG algorithm can recommend appropriate exercises for the targeted students.

Key words: knowledge, hierarchical graph, exercises recommendation, losing-rate matrix

摘要： 针对海量习题带来的信息过载导致学习针对性不强、效率不高等问题，提出了基于知识点层次图的个性化习题推荐算法（a personalized exercises Recommendation algorithm based on Knowledge Hierarchical Graph，ReKHG）。借鉴课程知识点体系结构的特点，构建了表征知识点层次关系的权重图，该权重图有效反映知识点间的层次关系。根据学生对知识点的掌握情况，在知识点层次图的基础上提出了一种个性化习题推荐算法。该算法通过更新学生-知识点失分率矩阵，获取学生掌握薄弱的知识点，以此实现习题推荐。实验结果表明，ReKHG算法能够针对性给学生推荐适合的习题。

关键词: 知识点, 层次图, 习题推荐, 失分率矩阵