With the wide availability of information over the internet and amount of its' potential users' makes it challenging for service provider to make such relevant information available to users in a fast and personalized manner nowadays, resulting in decrease in the web performance. One way to tackle with this challenging issue is to use prediction or recommendation technique, which can make users to discover future offerings. Web Usage Mining is a discipline, where the navigational access behavior of users' over the web is tracked and scrutinized. By collecting and analyzing this behavior of user activities, websites owner can easily identify the web access patterns of its users'. The mining of user access logs could help to improve the performance of search engines, since users have a specific goal when searching for information. In this paper, an approach is proposed to get more accurate prediction results by using both bi-clustering and greedy search method. The predictions of users' future access requests will be made by this can improve the accuracy of results and will helps in order to reduce the search time.

Future Requests Prediction Pattern Discovery Users' Navigational Behavior Web Usage Mining.