基于改进型协同过滤的网络学习资源推荐算法  

Recommendation Algorithm for E-learning Resources Based on Improved Collaborative Filtering

作　　者：王根生[1,2,3] 袁红林 黄学坚[2] 闵潞 WANG Gen-sheng;YUAN Hong-lin;HUANG Xue-jian;MIN Lu(School of International Trade and Economics,Jiangxi University of Finance and Economics,Nanchang 330013,China;Computer Practice Teaching Center,Jiangxi University of Finance and Economics,Nanchang 330013,China;School of Humanities,Jiangxi University of Finance and Economics,Nanchang 330013,China)

机构地区：[1]江西财经大学国际经贸学院,南昌330013 [2]江西财经大学计算机实践教学中心,南昌330013 [3]江西财经大学人文学院,南昌330013

出　　处：《小型微型计算机系统》2021年第5期940-945,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(71461012,72061015)资助;江西省教育厅科技基金项目(GJJ181550)资助;教育部科技发展中心产学研创新基金“智融兴教”项目(2018A01012)资助.

摘　　要：面对网络学习资源的信息过载问题,如何根据用户的偏好推荐其感兴趣的学习资源是网络教育智能化的关键应用.协同过滤推荐算法无需构建资源的特征描述,经常应用于形式多样的网络学习资源推荐,但传统协同过滤推荐算法具有评分矩阵稀疏和冷启动问题.针对这两个问题,提出基于改进型协同过滤的网络学习资源个性化推荐算法.该算法首先将用户对资源的学习行为转化成用户对资源的评分,缓解评分阵稀疏问题;其次引入用户初始化标签改进用户的相似度计算,解决新用户的冷启动问题;最后采用均方根误差(RMSE)进行推荐算法预测准确度衡量.实验结果表明,该改进算法提升了个性化资源推荐效果.Facing the problem of information overload of online learning resources,how to recommend the learning resources that users are interested in according to their preferences is the key application of network education intelligence.Collaborative filtering recommendation algorithm does not need to construct the feature description of resources,and is often used in various forms of network learning resource recommendation.However,the traditional collaborative filtering recommendation algorithm has the problems of sparse score matrix and cold start.Therefore,a personalized recommendation algorithm based on improved collaborative filtering is proposed.Firstly,the learning behavior of users on resources is transformed into users′rating of resources to alleviate the problem of sparse scoring matrix;Secondly,the user initialization tag is used to improve the user similarity calculation to solve the cold start problem of new users.Finally,RMSE is used to measure the prediction accuracy of recommendation algorithm.The experimental results show that the improved algorithm improves the personalized resource recommendation effect.

关 键 词：协同过滤 个性化推荐 网络学习资源 学习行为 用户初始化标签 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]