计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (13): 111-117.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

基于云计算技术的个性化推荐系统

应  毅1,2,刘亚军3,陈  诚4   

  1. 1.三江学院 软件学院,南京 210012
    2.三江学院 计算机科学与工程学院,南京 210012
    3.东南大学 计算机科学与工程学院,南京 210096
    4.易立方教育科技有限公司,南京 210012
  • 出版日期:2015-07-01 发布日期:2015-06-30

Personalization recommender system based on cloud-computing technology

YING Yi1,2, LIU Yajun3, CHEN Cheng4   

  1. 1.School of Software, Sanjiang University, Nanjing 210012, China
    2.College of Computer Science and Technology, Sanjiang University, Nanjing 210012, China
    3.School of Computer Science and Engineering, Southeast University, Nanjing 210096, China
    4.eCube Education Technology Co., Ltd., Nanjing 210012, China
  • Online:2015-07-01 Published:2015-06-30

摘要: 传统的协同过滤推荐技术在大数据环境下存在一定的不足。针对该问题,提出了一种基于云计算技术的个性化推荐方法:将大数据集和推荐计算分解到多台计算机上并行处理。在对经典ItemCF算法MapReduce化后,建立了一个基于Hadoop开源框架的并行推荐引擎,并通过在已商用的英语训练平台上进行学习推荐工作验证了该系统的有效性。实验结果表明,在集群中使用云计算技术处理海量数据,可以大大提高推荐系统的可扩展性。

关键词: 推荐系统, 基于物品的协同过滤, MapReduce, ItemCF-MR算法, 学习资源推荐

Abstract: Traditional collaborative filtering recommendation technology works poor under the environment of bigdata. To solve this problem, a personalized recommendation method based on cloud-computing technology is proposed. In this method, the large dataset and recommended calculation will be decomposed into multiple computers for parallel processing. It uses the open source framework Hadoop to establish a parallel recommendation engine on the basis of the classical ItemCF algorithm with MapReduce technology. The effectiveness of this system has already been verified on English training platforms by recommending learning resources. Experimental results indicate that the use of cloud-computing technology in the cluster to process massive data can significantly improve the scalability of Recommender Systems.

Key words: recommender systems, item-based collaborative filtering, MapReduce, ItemCF-MR algorithm, learning resources recommend