Semantic based Web Prefetching using Decision Tree Induction

Access delays on the web are the major issues, users experiencing these days. Web Prefetching has provided tremendous results in decreasing this access delay. Existing methods are based on URL´s of the web objects, user access. Use of domain ontology with the semantics can make this process easier and more effective. Semantics basically governs the relationships among objects of interest for the current scenario. Earlier methods also used to prefetch web pages, which were never being accessed. This paper provides a semantic based Prefetching scheme for the web browsers to overcome the limitations of existing systems. Semantic based prefetching uses anchor text as a base for finding out patterns. Anchor text is present in the hyperlinks of web page. With the help of anchor text, ranking of web page that will be received by search engine can also be determined. This scheme applies Decision Tree Induction for computing the probability of the anchor text and finding out patterns to be prefetched. SPRINT is used in the proposed framework as the decision tree induction method. For improving the performance, pruning is also applied on the Decision Tree Induction. The concept is elaborated with the help of example of mining mobile search engine. The experimental results are performed in Matlab 5.0.