In this paper, we propose to use database technology to improve performance of web proxy servers. We view the cache at a proxy server as a web warehouse with data organized in a hierarchical model, similar to data organization in database systems. The hierarchical model consists of physical pages, logical pages and topics, nding to different abstraction level. Based on this model, we then develop searching and topic navigation facilities for easily finding relevant contents in cache. By defining priorities on each abstraction level, the cache manager can make replacement decisions hierarchically based on the topics, logical pages and physical pages. As a result, the cache can always keep the most relevant, popular and high quality contents. We verify the proposed approach by developing a content-aware caching scheme, namely LRU-SP+. We evaluate our scheme in terms of hit ratio (HR) and profit ratio (PR) which differentiate topics of different pages. The results show that LRU-SP+ ge...
Kai Cheng, Yahiko Kambayashi, Mukesh K. Mohania