初始聚类中心优化的K-均值项目聚类推荐算法  

K-means item clustering recommendation algorithm with initial clustering center optimized

作　　者：胡旭[1] 鲁汉榕[1] 陈新[1] 周国安[1] 

机构地区：[1]空军预警学院

出　　处：《空军预警学院学报》2014年第3期203-207,共5页Journal of Air Force Early Warning Academy

摘　　要：针对协同过滤推荐系统存在的数据稀疏性和扩展性差问题,提出了初始聚类中心优化的K-均值项目聚类推荐算法.该算法首先采用SlopeOne方法对评分矩阵预测填充来缓解数据稀疏性,然后采用初始聚类中心优化的K-均值算法对项目进行聚类,将相似度高的项目聚到同一个类中,最后根据目标项目所在的聚类搜索其最近邻并产生推荐.实验结果表明,该算法有效改善了数据的稀疏性和扩展性,提高了推荐质量.In view of the problem of data sparsity and poor extensibility in collaborative filtering recommen-dation system, a K-means item clustering recommendation algorithm with initial clustering center optimized is proposed in this paper. This algorithm firstly fills the rating matrix by using the method of SlopeOne to relieve the data sparsity, and then clusters the items by using the K-means algorithm with initial clustering center optimized to cluster those items with high similarity into one category, and finally, search the nearest neighbor in terms of the cluster where the target item is located, and generate the recommendation resul. Experimental results show that this algorithm can improve the data sparsity and its extensibility effectively, bettering the recommendation quality.

关 键 词：协同过滤推荐 初始聚类中心优化 K-均值聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] O235[自动化与计算机技术—计算机科学与技术]