北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2009, Vol. 32 ›› Issue (6): 42-46.doi: 10.13190/jbupt.200906.42.zhangl

• 论文 • 上一篇    下一篇

基于BP神经网络的协作过滤推荐算法

张磊;陈俊亮;孟祥武;沈筱彦;段锟   

  1. (北京邮电大学 网络与交换技术国家重点实验室, 北京 100876)
  • 收稿日期:2009-04-07 修回日期:2009-05-22 出版日期:2009-12-28 发布日期:2009-12-28
  • 通讯作者: 张磊

BP Neural Networks-Based Collaborative Filtering Recommendation Algorithm

ZHANG Lei;CHEN Jun-liang;MENG Xiang-wu;SHEN Xiao-yan;DUAN Kun   

  1. (State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China)
  • Received:2009-04-07 Revised:2009-05-22 Online:2009-12-28 Published:2009-12-28
  • Contact: ZHANG Lei

摘要:

研究、探讨了协同推荐问题,提出了一种基于两层面的多个后向传播(BP)神经网络的协作过滤推荐算法(TMNN-CFRA). 两层面的多个BP神经网络协同工作,高层面BP网反向误差传播直至低层面多个人工神经网络(ANN)进行网络权值修正,以此为基础,借助用户评价等特征前向给出项目推荐. 标准评测集Movielens上的实验评测表明了TMNN-CFRA的可行性和有效性.

关键词: BP神经网络, 项目推荐, 协作过滤

Abstract:

A novel two-level multiple neural networks-based collaborative filtering recommendation algorithm (TMNN-CFA) for rating prediction is presented. By cooperating the multiple back propagation (BP) networks together, the higher layer neural network propagates conversely the output deviation until to the lower layer neural networks to amend the network weights and based on which, item recommendation is accomplished in the forward process of two layers networks relying on the factors such as ratings, etc. Experiment results on the standard Movielens show that TMNN-CFA method is effective and feasible for item recommendation.

Key words: back propagation neural networks, item recommendation, collaborative filtering