一种基于云模型数据填充的算法  

A Data Filling Algorithm Based on Cloud Model

作　　者：余志虎[1] 戚玉峰[1] 

机构地区：[1]南京邮电大学计算机学院,江苏南京210003

出　　处：《计算机技术与发展》2010年第12期34-37,共4页Computer Technology and Development

基　　金：国家863计划(2007AA701302;2009AA701202)

摘　　要：协同过滤推荐技术是现如今电子商务系统中最重要的技术之一。针对目前协同过滤推荐技术中存在的因数据极度稀疏而导致相似性度量不准,推荐质量严重受到影响的问题,利用云模型在定性知识表示及定性、定量知识转换时的作用,提出一种基于云模型的数据填充算法,它利用相似用户计算目标用户评分缺失项。利用经典实验数据进行验证比较,结果表明,即使在用户评分数据极端稀疏的情况下,利用此算法对数据进行填充之后,再采用传统的协同过滤推荐算法能取得较理想的推荐质量,从一定程度上解决了推荐系统中普遍存在的稀疏性问题。The collaborative filtering is one of the most important technologies in current E-commerce system,in the view of data in collaborative filtering technology are extremely sparse resulting in bad similarity measure and recommend poor quality,using a cloud model＇action in qualitative knowledge representation and the role of conversion among qualitative and quantitatve knowledge,proposed a data filling algorithm based on cloud model,and using the classical experimental data to validate and compare,it calculate user score missing items by using similar user.The result shows,even if the user＇s rating data is extremely sparse,it can get better recommendation quality by filling data with the algorithm and adopting traditional collaborative filtering algorithm,to some extent it can solve common sparse problems in recommended system.

关 键 词：云模型 数据稀疏 数据填充 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]