融合用户多维度偏好的慕课推荐研究  

A Study on Mooc Recommendation IncorporatingMultidimensional User Preferences

作　　者：邓伟伟 陈菁菁 陈寒 奉国和[1] DENG Wei-wei;CHEN Jing-jing;CHEN Han;FENG Guo-he(School of Economics and Management,South China Normal University,Guangzhou 510006,China;Department of Teacher Education,South China Normal University,Guangzhou 510631,China)

机构地区：[1]华南师范大学经济与管理学院,广州510006 [2]华南师范大学教师教育学部,广州510631

出　　处：《大学图书情报学刊》2025年第3期80-88,共9页Journal of Academic Library and Information Science

基　　金：国家自然科学基金青年项目“面向在线开放课程的捆绑推荐方法研究”(72301112);广东省自然科学基金面上项目“基于用户意图识别和学习路径规划的在线开放课程捆绑推荐方法研究”(2024A1515011842);广东省基础与应用基础研究基金省市联合基金“基于教学风格分析和教学质量评价的网络课程推荐方法研究”(2022A1515110677)。

摘　　要：慕课资源的快速增长给用户选择合适课程带来了信息过载问题。文章针对这一问题,提出有效的慕课推荐方法,以期降低用户的搜索成本,提高其学习体验和效果。首先,利用特征提取技术从课程、教师及学校介绍文本中提取多维度特征;然后,基于用户选课历史记录及其对应的多维度特征提取用户的多维度偏好;最后,将目标用户多维度偏好和候选课程对应的多维度特征输入极端梯度提升模型预测目标用户对候选课程的偏好,并根据预测偏好高低生成课程推荐列表。实验结果表明,融合用户多维度偏好的慕课推荐模型能够显著改善慕课推荐效果,且特征重要性和推荐案例分析进一步揭示了多维度特征在慕课推荐中的重要作用。The rapid growth of Mooc resources brings the problem of information overload to users in choosing appropriate courses.The article addresses this problem and proposes an effective Mooc recommendation method,with a view to reducing users'search cost and improving their learning experience and effect.First,we use feature extraction to extract multidimensional features from the text of course,teacher and school introduction;then,we extract users'multidimensional preferences based on users'historical course selection records and their corresponding multidimensional features;finally,we input target users'multidimensional preferences and multidimensional features corresponding to candidate courses into the extreme gradient boosting model to predict the target user's preference of candidate courses and generate a list of course recommendations based on the predicted preference level.The experimental results show that the Mooc recommendation model incorporating users'multidimensional preferences can significantly improve the effect of catechism course recommendation,and the importance of features and recommendation case analysis further reveal the important role of multidimensional features in catechism course recommendation.

关 键 词：多维度偏好 个性化推荐 慕课 机器学习 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]