江苏省科技期刊可视化阅读平台

导航条

《无线互联科技》杂志社 ›› 2025, Vol. 22 ›› Issue (1): 94-97.

• 研究创新 • 上一篇    下一篇

基于改进协同过滤的在线学习资源快捷推荐研究

高伟1,2, 郭丹1, 戴仁俊1   

  1. 1.江苏理工学院 计算机工程学院,江苏 常州 213001;
    2.扬州大学 教育科学学院,江苏 扬州 225000
  • 出版日期:2025-01-10 发布日期:2025-03-21
  • 作者简介:高伟(1989— ),男,讲师,硕士;研究方向:教育管理,数字媒体技术。
  • 基金资助:
    2020年度高校哲学社会科学研究一般项目;项目名称:基于MOOC平台的学习绩效研究与评价;项目编号:2020SJA1173。2022年度江苏理工学院教学改革项目;项目名称:基于模块化理念的在线教学视频资源库建设实践——以“摄影摄像技术”课程为例;项目编号:11610312305。江苏理工学院教改课题;项目名称:基于“互联网+教育”的在线翻转课堂教学模式设计与实践——以“现代教育技术”课程为例;项目编号:11610312120。

Research on quick recommendation of online learning resources based on improved collaborative filtering

GAO Wei1,2, GUO Dan1 , DAI Renjun1   

  1. 1. School of Computer Engineering, Jiangsu University of Technology, Changzhou 213001, China;
    2. School of Educational Science, Yangzhou University, Yangzhou 225000, China
  • Online:2025-01-10 Published:2025-03-21

PDF

232

摘要： 文章针对当前在线学习资源快捷推荐矩阵一般为单学科目标推荐,覆盖范围较小,导致最终得出的RMSE值增大问题,提出了一种基于改进协同过滤的在线学习资源快捷推荐方法。根据当前的资源推荐需求,先进行资源特征提取及标签化处理,采用多层级的方式,扩大覆盖范围,设计多层级的资源推荐矩阵。在此基础之上,构建改进协同过滤在线学习资源快捷推荐模型,采用自适应推荐评估的方式来实现处理。测试结果表明:对比Spark平台并行化谱聚类算法在线学习资源推荐方法、深度学习的个性化学习资源推荐方法,针对200人、400人、600人、800人和1200人5个小组,此次设计的改进协同过滤在线学习资源快捷推荐方法,在不同规模用户组中均能获得较低的最终RMSE值,这说明资源推荐的效率和精度得到了明显提高,具有实际的应用价值。

关键词: 改进协同, 过滤处理, 在线学习, 快捷推荐, 资源整合

Abstract: In view of the current fast recommendation matrix of online learning resources, the coverage is generally a single subject target, resulting in the increasing RMSE value, this paper proposes a quick recommendation method of online learning resources based on improved collaborative filtering. According to the current resource recommendation requirements, the resource feature extraction and labeling processing are conducted first, and the multi-level method is adopted to expand the coverage range and design the multi-level resource recommendation matrix. On this basis, the quick recommendation model of improved collaborative filtering online learning resources is constructed, and the adaptive recommendation evaluation method is adopted to realize the processing. Test results show that: compared to the Spark platform parallelized spectrum clustering algorithm online learning resources recommendation method, deep learning resources recommendation method, for 200,400,600,800 and 1200 five groups, the design of the improvement of collaborative filtering online learning resources quick recommendation method in different scale user group can get low final RMSE value is relatively small, that the efficiency and accuracy of resource recommendation is also improved, has the practical application value.

Key words: improve collaboration, filtering processing, online learning, quick recommendation, resource integration

中图分类号: