摘要
基于内容过滤和协同过滤是两大最为经典的推荐算法,但基于内容过滤存在新用户问题,没有考虑用户兴趣变化对推荐质量的影响,协同过滤则面临严峻的数据稀疏性和冷启动的挑战.针对这些,提出混合推荐算法:基于非线性逐步遗忘函数建立用户兴趣模型,预测用户未评价商品评分;引入"领域最近邻"处理方法查找目标用户的最近邻,预测未评价商品评分,以此为基础做出推荐.实验结果表明,本文方法能有效提高推荐质量.
Content-based filtering and collaborative filtering are the two most classical algorithms in recommendation system. However, there is the new customer problem in content filtering which does not consider the influences of users＇interests drifting on recommendation quality. And collaborative filtering faces severe challenges of data sparisity and cold start. To solve these problems, a hybrid recommendation algorithm is proposed in this paper. First,the paper builds a Customer Interests Model （CIM） based on the Forgetting Curve to predict the unevaluated rating ; then introduces the processing method of ＂domain nearest neighbor＂ to find the nearest neighbors for target users to predict the unevaluated rating; and finally, makes the recommendation. The experimental results show that the proposed method can improve the recommendation quality effectively.
出处
《管理科学学报》
CSSCI
北大核心
2012年第5期55-64,共10页
Journal of Management Sciences in China
基金
国家自然科学基金资助项目(70801026)
教育部人文社会科学基金资助项目(07JA630054)
教育部博士点基金新教师资助项目(200805321007)
关键词
混合推荐
非线性逐步遗忘
领域最近邻
hybrid recommendation
non-lineal gradual forgetting
domain nearest neighbor
作者简介
朱国玮（1978-），男，湖南长沙人，博士，副教授．Email：gwzhu@163．com
