Web Page Recommendation Using Efficient Weight Based Prediction System

Demands for more and more extraction of knowledge from the web data is increasing these days. This web data is sometimes available and sometimes hidden from the normal web user. In this regard, we studied the web usage pattern analysis. Usage mining is a kind of hidden web pattern analysis. Web usage data analysis is used in various applications such as user data recommendations, web mastering, web administrating, web pre-fetching and caching. The primary focus of this thesis work is to develop a web recommendation system. The proposed web recommendation system focuses on some relevant concepts such as behavioral analysis of user access patterns, personalization of data and predictive modeling. Proposed work gives the result obtained after study of a traditional recommendation system which is updated with some new additional parameter such as web usages data personalization, user navigational frequency analysis, session wise data analysis and time stamp data analysis. Additionally present work has also been concerned with predicting rarely accessed patterns. For this new weighted prediction technique is proposed in order to get more accurate user patterns. This weighted prediction technique includes isolation of the user personalization, reduced complexity and less resource consumption. This is observed that as the current access pattern or sequence increases, the prediction becomes more accurate and closer to actual page prediction. This prediction mechanism will greatly enhance performance and efficiency of proposed system.