摘要
协同过滤推荐是目前个性化推荐系统中使用最为广泛的方法.然而,传统协同过滤推荐一方面仅根据用户对项目的评分来判断用户之间是否存在共同喜好具有一定的片面性,因而降低了近邻搜索的质量;另一方面忽略了不同情境对用户偏好影响的差异性,进而影响了个性化推荐的效果.为此,提出一种基于情境化用户偏好的协同过滤推荐模型.首先,在模型中采用信息熵理论分析不同情境对用户偏好产生影响的重要程度,并结合用户-商品评分和用户对商品属性的偏好来搜索近邻用户;在此基础上,将情境重要度的权重引入到协同过滤推荐的生成过程中进而产生推荐结果.通过MovieLens数据集对该模型和其它两种协同过滤推荐进行比较的结果表明:本模型具有较低的平均误差,进而表明了考虑情境化用户偏好的协同过滤可明显改善个性化推荐的质量.
The collaborative filtering （CF） is the most widely used recommendation technology in per- sonalized recommender systems. However, it has one-sidedness in determining the common preferences between users because the traditional CF calculates user similarities only according to user-item ratings, and thus reduces the quality of searching neighbor. Besides, many context-based CF methods ignore the differences of the effect of varieties contexts to user preferences, and thus affect the recommendation effec- tiveness, To address these problems, this paper proposes an improve CF model based on contextualized user preferences. Firstly, the proposed model analyses the importance of the effect of different contexts to user preferences based on the theory of information entropy, and searches nearest neighbors accord- ing to user-commodity ratings and user preferences to commodity attributes. And then, the weight of context importance is introduced in the process of recommendation generation to obtain the recommenda- tion results. To evaluate the performance of the proposed model, a set of the experiments on MovieLens dataset are conducted, and the results show that the proposed model has low MAE value than other CF methods, thus it enhances the prediction accuracy and improves the quality of context-based personalized recommendation.
作者
吕苗
金淳
韩庆平
LU Miao JIN Chun HAN Jim C(Institute of Systems Engineering, Dalian University of Technology, Dalian 116024, China Department of Information Technology and Operations Management, Florida Atlantic University, FL 33431, USA)
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2016年第12期3244-3254,共11页
Systems Engineering-Theory & Practice
基金
国家自然科学基金重大项目(70890080
70890083)~~
关键词
协同过滤
情境
用户偏好
信息熵
个性化推荐
collaborative filtering
context
user preference
information entropy
personalized recommen- dation
