基于改进图神经网络和用户偏好聚类的个性化学习资源推荐算法  

Personalized learning resource recommendation algorithm based on improved graph neural network and user preference clustering

作　　者：王慧[1] 孙德红[1] WANG Hui;SUN Dehong(School of Information Management,Minnan University of Science and Technology,Shishi 362700,China)

机构地区：[1]闽南理工学院信息管理学院,福建石狮362700

出　　处：《黑龙江工程学院学报》2022年第6期30-34,共5页Journal of Heilongjiang Institute of Technology

基　　金：2019年福建省中青年教师教育科研项目(JAT190860)。

摘　　要：由于学习资源中知识点构成的复杂性以及用户需求的差异性,导致最终的推荐结果存在较大差异,为此,提出基于改进图神经网络和用户偏好聚类的个性化学习资源推荐算法。从交流互动、自主学习意识以及学习能力3个方面分析用户的偏好,在图神经网络中引入元数据概念,建立以知识点为基础的学习资源本体,实现资源之间的关联关系,对用户偏好和资源本体双重聚类后,匹配类之间的拟合关系,确定推荐内容。通过实验测试验证该设计方法推荐内容的精确率为60.96%,召回率为65.42%,F综合评价指标为62.57%,说明该方法具有良好的性能。Due to the complexity of knowledge points in learning resources and the differences of user needs, the final recommendation results are quite different. Therefore, a personalized learning resource recommendation algorithm based on improved graph neural network and user preference clustering is proposed. Taking communication and interaction, self-learning consciousness and learning ability as the starting point, this paper analyzes the preferences of users. The concept of metadata is introduced into the graph neural network to construct the learning resource ontology based on knowledge points, so as to realize the correlation between resources. After double clustering of user preferences and resource ontology, the recommended content is determined, and the personalized learning resource recommendation is realized. The experimental test shows that the accuracy rate of the recommended content of this design method is 60.96%, the recall rate is 65.42%, and the Fcomprehensive evaluation index is 62.57%, which proves that this method has superior performance.

关 键 词：改进图神经网络 用户偏好聚类 个性化 学习资源 推荐算法 双重聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]