Vol. 3,No. 9, September 2013
Author(s): Zeinab Khademali, Nima Attarzadeh, Mohammad Mehdi LotfiNejad
Abstract: In this paper we try to classify navigation patterns of web users automatically; therefore, a new method is presented in order to classify the user’s navigation patterns and predict the user’s future requirements. This method is based on the mining of web server logs. Thus, in order to build user’s profile, a new method is introduced that, by registering user’s setting and similarity measure of active user to neighboring users, constructs indices implicitly and brings them up to date based on created changes. In effect, we improve the performance of recommender engine by using navigation patterns of user and clustering similar users. Furthermore, we test the precision for different inputs in the model simulated based on the neural network and we determine that if, in registering user’s profile, in addition to behavior history, current session is focused on as well, recommender engine will offer better results. This method that is based on user’s navigation patterns is capable of offering the results from recommender engines based on user’s requirement and interest. Advantage is evaluated based on two responsibilities: classification and prediction. The system has reached classification precision close to and prediction precision of about .This method can help web personalization and website better organization.
International Association for Academians is an International institute with regional headquarters in Canada.