基于项目属性和云填充协同过滤推荐算法.docVIP

基于项目属性和云填充协同过滤推荐算法 　　摘要:传统协同过滤推荐算法中经常因用户评分矩阵极端稀疏而导致相似性度量方法不准,推荐质量不高,针对这一问题,提出一种基于项目属性和云填充的协同过滤推荐算法。利用云模型对用户评分矩阵进行填充,在填充矩阵基础上,利用传统的相似性计算方法得到项目之间的评分相似性,同时结合项目属性,计算项目的属性相似性,通过加权因子得到项目的最终相似性,从而形成一种新的相似性度量方法。实验结果表明, 提出的算法可有效解决传统方法中由于数据稀疏所导致的相似性度量不准确的问题, 并显著地提高了算法的推荐精度。 　　关键词:协同过滤;稀疏数据;云填充;评分相似性;属性相似性;相似性度量 　　中图分类号: TP18;TP301.6文献标志码:A 　　Collaborative filtering recommendation algorithm 　　based on item attribute and cloud model filling 　　英文作者名SUN Jin-gang*, AI Li-rong 　　英文地址(School of Computer Science and Technology, Northwestern Polytechnical University, Xi?an Shaanxi 710129, China) 　　Abstract: The user rating data in traditional collaborative filtering recommendation algorithm are extremely sparse, which results in bad similarity measurement and poor recommendation quality. In view of this problem, this paper presented an improved collaborative filtering algorithm, which was based on item attribute and cloud model filling. The algorithm proposed a new similarity measurement method, using the data filling based on cloud model and the similarity of the item?s attributes. The new method computed the rating similarity by using the traditional similarity measurement on the basis of the filling matrix and computed the attributing similarity by using item?s attributes, then got the last similarity by using weighting factor. The experimental results show that this method can efficiently solve the problem of similarity measurement inaccuracy caused by the extreme sparsity of user rating data, and provide better recommendation results than traditional collaborative filtering algorithms. 　　Key words: collaborative filtering; sparse data; cloud model filling; rating similarity; attributing similarity; similarity measure 　　0引言 　　Internet 技术的应用普及和现代电子商务的迅猛发展使得互联网中的资源数量呈指数增长态势,从而出现了所谓的“信息???炸”和“信息过载”现象,推荐系统[1]作为一种信息过滤的重要手段, 是当前解决上述问题的有效方法之一。目前,几乎所有大型的电子商务系统,如淘宝、当当、Amazon、eBay等都不同程度地使用了各种形式的推荐系统。 　　随着推荐系统规模的扩大,用户评分数据出现极端稀疏性[2],导致算法的推荐质量降低。为了解决稀疏性问题,一些学者提出了算法改进措施:文献[3]提出基于云模