计算机科学 ›› 2017, Vol. 44 ›› Issue (Z6): 438-441.doi: 10.11896/j.issn.1002-137X.2017.6A.098

• 大数据与数据挖掘 • 上一篇    下一篇

大数据环境下基于概率矩阵分解的个性化推荐

田贤忠,沈杰   

  1. 浙江工业大学计算机科学与技术学院 杭州310023,浙江工业大学计算机科学与技术学院 杭州310023
  • 出版日期:2017-12-01 发布日期:2018-12-01
  • 基金资助:
    本文受国家自然科学基金项目(61672465),浙江省自然科学基金项目(LY15F020027)资助

Personalized Recommendation Based on Probabilistic Matrix Factorization in Big Data Environment

TIAN Xian-zhong and SHEN Jie   

  • Online:2017-12-01 Published:2018-12-01

摘要: 概率矩阵分解是近几年广泛应用的协同过滤推荐方法。针对如何利用矩阵分解技术提高推荐质量以及在大数据环境下如何突破计算时间、计算资源瓶颈等问题进行研究,提出了Improved Probabilistic Matrix Factorization(IPMF)融入邻居信息的概率矩阵分解算法,并且提出了parallel-IPMF (p-IPMF)算法来解决融入邻居信息后计算复杂度高和难以并行化等问题。 在MapReduce并行计算框架下将p-IPMF算法加以实现,并在真实数据集上进行验证。实验结果表明,所提算法能有效提高推荐质量并缩短计算时间。

关键词: 推荐算法,概率矩阵分解,大数据,MapReduce

Abstract: Probabilistic matrix factorization is a type of collaborative filtering algorithm which is widely used in recent years.Based on the problem of how to use matrix factorization technology to improve the recommendation quality and how to breakthrough the limitation of calculation time and resource in big data environment,we introduced an improved probabilistic matrix factorization algorithm which integrates neighbor information and introduced parallel-IPMF,overcoming the problem of high calculation complex and the problem of parallelization.We used the real dataset to implement our algorithm on the MapReduce parallel computation framework.The experiment results show that our algorithm can improve the recommendation quality and reduce the computation time.

Key words: Recommendation algorithm,Probabilistic matrix factorization,Big data,MapReduce

[1] ADOMAVICIUS G,TUZHILIN A.Toward the Next Generation of Recommender Systems:A Survey of the State-of-the-Art and Possible Extensions[J].IEEE Transactions on Knowledge and Data Engineering,2005,7(6):734-749.
[2] SALAKHUTDINOV R,MNIH A.Probabilistic Matrix Factorization[M]∥Advances in neural information processing systems,NIPS’08.Cambridge,Massachusetts,USA,MIT Press,2008:1257-1264.
[3] ZHANG Z J,LIU H.Social Recommendation Model Combining Trust Propagation and Sequential Behaviors[C]∥Applied Intelligence.2015.
[4] LIU Q,WANG C W,XU C F.A modified PMF modelincorporating implicit item associations[C]∥Proceedings of 24th International Conference on Tools with Artificial Intelligence.Athens,Greece,2012.
[5] CHAKROUN I,Haber T,AA T V.Exploring Parallel Implementations of the Bayesian Probabilistic Matrix Factorization[C]∥Parallel,Distributed,and Network-Based Processing (PDP).2016.
[6] GEMULLA R,HAAS P J,NIJKAMP E,et al.Large-Scale matrix factorization with distributed stochastic gradient descent[C]∥Proc.of the 17th ACM SIGKDD Int’l Conf.on Know-ledge Discovery and Data Mining.ACM Press,2011:69-77.
[7] RECHT B,R C.Parallel stochastic gradient algorithms forlarge-scale matrix completion[J].Mathematical Programming Computation,2013,5(2):201-226.
[8] 印鉴,王智圣,李琪,等.基于大规模隐式反馈的个性化推荐[J].软件学报,2014,5(9):1953-1966.
[9] Lmmel R.Google’s MapReduce programming model-revisited[J].Science of Computer Programming,2007,68(3):208-237.
[10] 王全民,苗雨,何明,等.基于矩阵分解的协同过滤算法的并行化研究[J].计算机技术与发展,2015,5(2):55-59.
[11] Cloudera.Cloudera[EB/OL].http://www.cloudera.com.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!