The World Wide Web is a distributed internet system, which provides a many type of services and facilities for users. It increase over the past few year at a very rapidly rate, due to which the amount of traffic over the internet is increasing. As a result, the network performance has now become very slow. The solution of this is to reduce the response time perceived by users. The web prefetching is an effective user prediction technique that extracts useful knowledge from user request sequence. It makes a prediction of the web pages that the user is likely to request in the near future. Web prefetching is one of the effective solutions used to reduce web access latency and improve the quality of service. Many researchers had proposed various predictions based prefetching algorithm and model such as N-gram based prediction [10], PPM model [11], Dynamic prefetching technique [12], and other machine learning technique such as Matrix prefetching [13], Multi dimensional matrix prefetching model [4] and Prediction based model [9]. These techniques are low hit rate and byte hit rate. The proposed SVM regression used to predict the user’s future request for prefetching. This technique to improve system performance and increase hit rate and byte hit rate. Therefore our research is focus to apply SVM technique to the problem of user action for prediction on the web. In particular, we are able to predict the future web page that a user will select. Through the simulation, we found that our approach has quantitative measures such as hit rate and byte hit rate of accessed page.

Latency SVM regression Prediction Quantitative Web Prefetching