《计算机应用研究》|Application Research of Computers

耦合辅助信息的矩阵分解推荐模型

Matrix factorization recommendation model based on side information

免费全文下载 (已被下载 次)  
获取PDF全文
作者 蒋伟,秦志光
机构 电子科技大学 信息与软件工程学院,成都 610054
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)10-004-2900-07
DOI 10.19734/j.issn.1001-3695.2018.03.0206
摘要 近十年来,协同过滤(CF)推荐系统成功地为用户提供了个性化的产品和服务。然而,用户—物品矩阵的稀疏性、推荐精度不高等问题仍然是一个挑战。针对这些问题,在矩阵分解模型基础上,提出了耦合用户和物品辅助信息的矩阵分解混合协同过滤框架;然后,基于此框架又提出了耦合物品属性信息相似度(COS)的过滤模型。大规模真实数据集上的实验表明,该模型不但可以有效解决物品相似度度量问题,而且相比传统方法,尤其是在物品特征非常稀疏的情况下,推荐准确性得到了有效改进。
关键词 推荐系统; 混合协同过滤; 矩阵分解; 物品相似度; 耦合对象相似度; 辅助信息
基金项目 四川省科技计划资助项目(2015JY0178,2014GZ0109,2015KZ002,2015JY0030)
国家自然科学基金资助项目(61472064)
中央高校基本科研基金资助项目(ZYGX2014J051,ZYGX2014J066)
本文URL http://www.arocmag.com/article/01-2019-10-004.html
英文标题 Matrix factorization recommendation model based on side information
作者英文名 Jiang Wei, Qin Zhiguang
机构英文名 School of Information & Software Engineering,University of Electronic Science & Technology of China,Chengdu 610054,China
英文摘要 Collaborative filtering(CF) recommender systems have been used to provide users with personalized products and services successfully in the past decade. However, sparseness of user-item matrix and the low accuracy are still challenges. To solve these problems, this paper proposed an ensemble framework based on matrix factorization CF for integrating side information of users and items. Based on this framework, this paper proposed a hybrid CF model for integrating COS(coupled object similarity) of attribute information of items. Extensive experiments conducted over large-scale real-word datasets demonstrate that the proposed approach can effectively solve the problem of item similarity measurement. And compared with the traditional approaches, especially in the case of very sparse feature, the accuracy of the recommendation is improved effectively.
英文关键词 recommender system; hybrid collaborative filtering; matrix factorization; item similarity; coupled object similarity; side information
参考文献 查看稿件参考文献
 
收稿日期 2018/3/27
修回日期 2018/5/11
页码 2900-2906
中图分类号 TP181
文献标志码 A