摘要
针对数据稀疏性与推荐实时性的技术难题,在结合传统用户合作过滤推荐的基础上,提出了基于兴趣度向量模型的用户合作推荐机制.该方法合理利用了用户的人口统计信息,即用户提交给网站的注册信息,来辅助基于兴趣度向量模型的推荐方法,在提高预测精度的同时还可以解决推荐系统的新用户问题.采用MovieLens网站上提供的研究数据进行模拟推荐实验,通过对平均绝对偏差对比分析可得,兴趣向量模型的推荐方法在一定程度上比传统合作过滤算法有更高的推荐精度.
In order to solve the technological problems in the data sparsity and real time recommendation, a customer cooperation recommendation mechanism based on the interest degree vector model was proposed with integrating the traditional customer cooperation filtering technology. The proposed method utilized the demographic information of customers, the registration information submitted to the web site by customers, to assist the recommendation method based on interest degree vector model. It can improve the prediction precision and solve the new customer problem in the recommendation system. The simulation experiment was carded out using the data given on the MovieLens web site. The analysis of the mean absolute error shows that the method based on interest degree vector model has better recommendation precision than the traditional cooperation filtering algorithm.
出处
《沈阳工业大学学报》
EI
CAS
2009年第5期573-576,595,共5页
Journal of Shenyang University of Technology
基金
国家青年科学基金资助项目(60905054)
关键词
合作过滤
兴趣度向量模型
平均绝对偏差
人口统计信息
推荐系统
数据稀疏性
推荐机制
推荐方法
cooperation filtering
interest degree vector model
mean absolute error
demographic information
recommendafion system
data sparsity
recommendation mechanism
recommendation method