Web clustering based prefetching in high traffic environment

The need of minimizing the latency perceived by the user in fetching web objects without necessarily increasing the bandwidth has attracted several researchers in the recent years. Although web prefetching and caching is seen as a solution, proposed techniques do not consider the frequency of server idle time in their models. This paper therefore proposes a short time web prefetching framework based on clustering technique that can be effective in high traffic with low bandwidth environment where the server idle time is too minimal to fetch all users anticipated requests. Clusters from different user requests are used to perform an inter domain clustering that prioritizes the prefetching of web pages based on speed at which requests are received from each domain and the popularity of each page. Experimental results show an improvement in hit rate and precision over the classical clustering based prefetching technique when the server idle time is not enough to prefetch all clusters.