基于相似云与复合因素度量的个性化推荐算法  

Personalized Recommendation Algorithm Based on Similar Cloud and Measurement by Multifactor

作　　者：孙光明[1] 王硕[2] 李伟生[1] SUN Guang-ming WANG Shuo LI Wei-sheng(School of Computer and Information Technology, Beij ing Jiaotong University, Beijing 100004, China School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050035, China)

机构地区：[1]北京交通大学计算机与信息技术学院,北京100004 [2]河北科技大学信息科学与工程学院,石家庄050035

出　　处：《计算机科学》2016年第8期165-170,198,共7页Computer Science

基　　金：河北省高等学校科学技术研究重点项目(ZD2014061)资助

摘　　要：针对相似计算中评分数据的稀疏性、属性严格匹配与单因素度量的偶然性导致的近邻不准问题,提出基于相似云与复合因素度量的个性化推荐算法。该算法定义按类的项目评分云来预填充评分,提出融合类别、评分均值、评分频度、访问频度等多因素度量的项目兴趣度向量,通过云模型计算项目相似度,以按类预测其评分,并基于新的加权平均方法计算其最终评分值。实验结果表明,所提算法产生的近邻更准,推荐质量更高。Aiming at the problems of sparse data,occasionality that caused by attribute＇s strict matching and measuring with single rating in similarity computation, this paper presented a personalized recommendation algorithm based on similar cloud and measurement by multifactor. The algorithm defines a marking cloud classified by item to fill sparse matrix,and puts forward a feature vector of item＇s interest degree consisting of item category,mean score, frequency of rating and accessing,which contributes to calculating the item＇s similarity with cloud model. On this basis, this algo- rithm predicts item＇s score in different categories according to the nearest neighbors, and gets the item＇s finally score based on a new weighted average computing method. Experimental results show that the algorithm produces more accu- rate neighbors and better recommendation quality.

关 键 词：相似云 复合因素度量 相似度 加权平均 个性化算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]