协同过滤算法下公共计算机慕课资源推荐系统

Public computer MOOC resource recommendation system based on collaborative filtering algorithm

公共计算机慕课资源量较大,易受到冗余信息的干扰,导致学生需求和学习资源难以匹配,推荐误差较大。为此,文中设计了协同过滤算法下公共计算机慕课资源推荐系统。该系统使用DAGScheduleX调度引擎,利用控制台绑定不同类型集群调度推荐资源,在不同用户场景下分解请求源任务。首先通过特征服务平台,将个性化特征信息放入本地磁盘文件中,分析公共计算机慕课资源推荐列表中最大支持度项目集合,计算推荐资源估计参数置信区间,对冗余数据进行分词、去停用词预处理;然后参照用户对资源评分结果搜索最近邻居,通过协同过滤冗余信息获取推荐结果;最后根据利用计算信息相似性得到的评分结果向用户推荐慕课资源。系统测试结果表明,该系统能够提取到78%的公共计算机慕课资源,最大推荐误差为10%,能够为用户提供感兴趣资源。

Because of the large amount of MOOC resources,it is easy for the MOOC resource to be interfered by redundant information,which makes it difficult to match the needs of students with learning resources,so that the recommendation error is large.Therefore,the public computer MOOC resource recommendation system based on collaborative filtering algorithm is designed.The DAGScheduleX scheduling engine is used in the system to bind different types of cluster scheduling recommended resources by the console.The request source task is decomposed in different user scenarios.The personalized feature information is put into the local disk file by means of the feature service platform.The maximum support item set in the recommendation list of public computer MOOC resources is analyzed,the confidence intervals of the estimated parameters of recommended resources are calculated,and the redundant data are preprocessed by word segmentation and de⁃stopping.The nearest neighbor is searched according to the user′s resource score,and the recommendation result is obtained by collaborative filtering of redundant information.The similarity of information is calculated.The rating results are combined to recommend MOOC resources to users.The system test results show that the system can extract the public computer MOOC resources of 78%,the maximum recommendation error is 10%,and the system can provide users with interested resources.