• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2015, Vol. 37 ›› Issue (11): 2128-2133.

• 论文 • 上一篇    下一篇

基于协同过滤的Web缓存替换算法研究

吴俊龙,杨清   

  1. (湖南科技大学计算机科学与工程学院,湖南 湘潭 411201)
  • 收稿日期:2014-12-20 修回日期:2015-04-08 出版日期:2015-11-25 发布日期:2015-11-25
  • 基金资助:

    湖南省科技计划资助项目(2013FJ4050)

A web cache replacement algorithm based on collaborative filtering 

WU Junlong,YANG Qing   

  1. (School of Computer Science and Engineering,Hunan University of Science and Technology,Xiangtan 411201,China)
  • Received:2014-12-20 Revised:2015-04-08 Online:2015-11-25 Published:2015-11-25

摘要:

针对GDSF替换算法中对访问频率缺少预测的不足,提出了一种基于协同过滤的GDSF缓存替换算法(GDSFCF)。该算法考虑了Web对象之间相似性与用户访问时间间隔,运用协同过滤算法生成Web对象的预测访问频率,并采用齐普夫定律参数对GDSF算法的目标函数进行了改进。当需要进行缓存替换时,利用目标函数价值计算缓存空间中的每个Web对象缓存价值,将最小缓存价值的Web对象进行替换。仿真实验结果表明,该算法的命中率HR和字节命中率BHR都有较大提升。

关键词: Web缓存, 协同过滤, 替换算法

Abstract:

To address the shortcoming that the GDSF replacement algorithm cannot predict visit frequency, we propose a collaborative filtering based GDSF web cache replacement algorithm (GDSFCF). Considering the similarities among web objects and the time interval of users’ visits, we use the collaborative filtering technology to generate the predicted visit frequency of each web object, and adopts Zipflike parameters to modify the objective functions of the GDSF algorithm. When cache replacement is performed, the value of the objective functions are used to calculate the caching value of each web object, and the web object with the minimum value will be replaced in the first place. Simulation experiments show that the GDSF-CF algorithm has a higher hit rate and byte hit rate.

Key words: Web cache;collaborative filtering;replacement algorithm