基于混合模式的网络超市商品推荐方法  

Online Supermarket Goods Recommendation Method Based on Mixed Mode

作　　者：温廷新[1] 唐小龙[1] 马龙梅[1] 

机构地区：[1]辽宁工程技术大学工商管理学院,辽宁葫芦岛125105

出　　处：《现代情报》2013年第12期45-51,共7页Journal of Modern Information

基　　金：辽宁省突发事件应急管理的多元化IS体系设计(项目编号:LT2010048);山东省突发事件多元应急信息系统研究与构建(项目编号:ZR2010FL012)

摘　　要：在分析现有网络商品推荐算法的优缺点及主要问题的基础上,提出一种基于混合模式的网络超市商品推荐方法。其主要思想是:通过商品本体概念和属性构建商品子模型,采用基于内容的推荐算法填充用户——商品评分矩阵;依据用户背景信息、评分数据和查询关键字构建用户子模型,采用K均值算法进行用户聚类;利用基于用户的协同过滤产生推荐。实验表明,混合算法提供的推荐结果更加准确高效。Aftex analyzlng the advantages and disadvantages of the current recommendation algorithms and the main prob- lems of algorithms, the online supermarket goods recommendation method based on mixed mode had been put forward. In order to fill the user- rating- data matrix via adopting contem- based recommendation, this paper constructed commodity sub- model by using good ontology concepts and attributes of goods. In the same way, it constructed user sub - model by using lmelound infor- marion, rating data and query keywords of user to complete the k- means clustering. Then the results were generated by means of collaborative fiiterin~ algorithm. The experimental results suggested that hybrid algorithm revealed good performance.

关 键 词：个性化推荐 网络超市 混合模式 相似度 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]