Association rule based data mining agents for personalized web caching

  • Authors:
  • Sujaa Rani Mohan;E. K. Park;Yijie Han

  • Affiliations:
  • University of Missouri, Kansas City;University of Missouri, Kansas City;University of Missouri, Kansas City

  • Venue:
  • COMPSAC-W'05 Proceedings of the 29th annual international conference on Computer software and applications conference
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Proxy web caching is commonly implemented to decrease web access latency, internet bandwidth costs and origin web server load. We propose a transparent shareable proxy caching methodology in which the proxy caches maintain a continuously optimal performance with a significant improvement in the cache hit ratio without requiring any additional overhead at the client or at the routers. This approach allows caches' to be personalized based on the users' web access patterns. Our approach employs four light weight intelligent agents in a system of proxy caches which use classification algorithms, data mining techniques and performance monitors to study the client's web access patterns and configure caches to best serve the clients with available resources. These agents run with predictable run times with an optimal use of computer resources.