Semantic Web Prefetching Scheme using Naïve Bayes Classifier

Web prefetching is an effective technique to minimize user’s web access latency. Web page content provides useful information for making the predictions that is used to perform prefetching of web objects. In this paper we propose semantic prefetching scheme that uses anchor texts present in the web page to make effective predictions. The scheme applies Naïve Bayes Classifier for computing the probability values of anchor texts, based on which the predictions are generated and given as input to the prefetch unit. Predictions are dynamic and are based on the browsing pattern exhibited by the user in a session. The browsing pattern of user should be specific towards a particular topic of interest to achieve good hit rate. When user has long browsing sessions, predictions and prefetching are more effective and helps to increase the hit rate and minimize access latency.