计算机与现代化

• 网络与通信 • 上一篇    下一篇

基于增强关联规则挖掘的大型网站推荐系统

  

  1. 东莞理工学院计算机学院,广东东莞523808
  • 收稿日期:2016-03-29 出版日期:2016-10-15 发布日期:2016-10-14
  • 作者简介:邹裕(1983-),男,广东河源人,东莞理工学院计算机学院实验师,硕士,研究方向:计算机应用,移动互联网; 肖倩(1973-),女,湖南桃江人,高级实验师,硕士,研究方向:智能算法,计算机应用 ; 吴树荣(1977-),男,广东东莞人,实验师,硕士,研究方向:网络安全,智能算法。
  • 基金资助:
    广东省高校优秀青年创新人才培养计划项目(2012LYM0125)

Recommendation System for Large Website Based on Enhanced Association Rules Mining

  1. College of Computer Science and Technology, Dongguan University of Technology, Dongguan 523808, China
  • Received:2016-03-29 Online:2016-10-15 Published:2016-10-14

摘要:

针对大型网络中基于内容的推荐系统延迟较高、推荐满意度较低的问题,提出一种基于增强关联规则挖掘算法的高性能推荐系统。首先,通过消除会话中出现频率较低的内容将数据库最小化处理
;然后,为每个会话引入相关的计数器,使用该计数器统计相应会话的重复次数,代替传统推荐系统对重复会话的直接聚类,该改进策略不仅提高了推荐的时间效率,而且增强了推荐引擎的扩展能力。
基于真实数据的实验结果表明,本推荐系统具有较高的实时性与较高的推荐准确率。

关键词: 关联规则, 推荐系统, 实时性, 会话计数器, 推荐引擎

Abstract:

Aiming at the problem that the content based recommendation system has high delay and low degree of satisfaction of recommendation in the large web site, an enhanced
association rules mining based recommendation system for large web site is presented. Firstly, the items with low frequency occurrence are eliminated so that the database is
minimized; Then, a counter is introduced to each transaction, and the counter is used to count the number of replication of the transaction instead of the directly clustering
process of repeated transactions in the traditional recommendation system, this improved strategy not only improves the efficiency of recommendation but also the scalability of
the recommendation engine. Real data based experiments results show that the proposed system is real time and has high recommendation accuracy.

Key words: association rules, recommendation system, real time, transaction counter, recommendation engine

中图分类号: