电子学报 ›› 2016, Vol. 44 ›› Issue (7): 1581-1586.DOI: 10.3969/j.issn.0372-2112.2016.07.009

• 学术论文 • 上一篇    下一篇

基于协同矩阵分解的评分与信任联合预测

张维玉1,2, 吴斌1, 耿玉水2, 朱江1   

  1. 1. 北京邮电大学智能通信软件与多媒体北京市重点实验室, 北京 100876;
    2. 齐鲁工业大学信息学院, 山东济南 250353
  • 收稿日期:2015-02-06 修回日期:2016-01-20 出版日期:2016-07-25 发布日期:2016-07-25
  • 作者简介:张维玉 男,1978年6月生于山东兖州.现在北京邮电大学攻读博士学位,研究方向为社交网络分析、数据挖掘和机器学习.E-mail:zwy@bupt.edu.cn;吴斌 男,1969年11月生于湖南长沙.现为北京邮电大学博士生导师,研究方向为图数据挖掘、智能信息处理
  • 基金资助:

    国家973重点基础研究发展计划项目(No.2013CB329606);国家自然科学基金(No.71231002,No.61375058)

Joint Rating and Trust Prediction Based on Collective Matrix Factorization

ZHANG Wei-yu1,2, WU Bin1, GENG Yu-shui2, ZHU Jiang1   

  1. 1. Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, Beijing 100876, China;
    2. School of Information, Qilu University of Technology, Jinan, Shandong 250353, China
  • Received:2015-02-06 Revised:2016-01-20 Online:2016-07-25 Published:2016-07-25

摘要:

信息评分预测和信任预测是社交评价网络中的两大基本问题.为应对在提高两类基本问题预测准确性过程中遇到的评分数据与信任关系数据稀疏问题,本文提出了一种基于协同矩阵分解的信息评分与信任预测联合模型.该模型在将评分矩阵与信任关系矩阵进行协同分解时,既能保证被分解的两个矩阵分解过程共享用户潜在变量,又能兼顾两个矩阵分解过程中能够各自获得反映本领域知识相关性的表达.使用分解得到的多个相关低维潜在变量矩阵乘积即可做出评分与信任两个问题的预测.两个真实网络数据集上的实验验证了提出模型有效性和先进性.

关键词: 推荐算法, 信任预测, 概率矩阵分解, 社交推荐网络

Abstract:

Trust prediction and information item rating are two fundamental tasks for social rating network systems.In response to data sparsity of trust relation and rating information encountered when improving the accuracy of predicting the two basic problems,we present a joint model of rating and trust prediction based on collective matrix factorization.In our model,trust relation matrix and information rating matrix are factorized into latent features matrixes collectively.We can make full use of correspondence among users and information items by sharing latent user feature.Moreover,our model can capture the data dependent effect of trust domains and rating domain separately.By using those learned latent features matrixes multiplication,we can obtain predictions of trust and rating.Experimental results on two real network data demonstrate that our model is more accurate than other state-of-the-art methods.

Key words: recommendation algorithms, trust prediction, probabilistic matrix factorization, social rating networks

中图分类号: