基于改进的协同过滤算法的练习测试推荐系统  

Recommendation system based on improved collaborative filtering algorithm for exercise test

作　　者：徐琳[1] 杨姝[1] 

机构地区：[1]沈阳师范大学教育技术学院,沈阳110034

出　　处：《沈阳师范大学学报（自然科学版）》2015年第2期270-273,共4页Journal of Shenyang Normal University:Natural Science Edition

基　　金：辽宁省教育厅高等学校科学研究项目(w2014138)

摘　　要：在传统的协同过滤推荐算法的基础上,设计了一个基于改进的协同过滤算法的练习测试推荐系统。首先,根据学科、试题和学生的特点,有效的解决了矩阵稀疏和"冷启动"的问题;其次,使用机器学习中的K-means聚类算法对用户进行聚类,且初始聚类中心由Prim最小生成树算法确定,增加了聚类的稳定性;然后在每个聚类中搜索用户的最近邻居,缩小了计算用户之间相似度问题的规模;最后,通过实验将改进的算法与传统的算法进行了比较。实验结果表明,改进的算法提高了推荐系统的质量和准确度。On the basis of traditional collaborative filtering recommendation algorithm, this paper designs a recommendation system based on the improved collaborative filtering algorithm for the exercise test. First of all, according to the characteristics of the subjects, exam, and students, it effectively solves the sparse matrix and cold start problem; Secondly, it uses the K-means clustering algorithm of machine learning to cluster students, and find the initial clustering center by the prim minimum spanning tree algorithm, in order to increase the stability of clustering; Then this paper searches the users＇ nearest neighbors in each cluster, by way of reducing the scale of the problem that computing the similarity between users; Finally through experiment, it compares the improved algorithm with the traditional algorithm. The results show that the improved algorithm improve the quality and accuracy of recommendation system.

关 键 词：协同过滤 推荐系统 成绩评价 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]