چكيده به لاتين
Nowadays, web-based services like E-Commerce, E-Learning and E-Banking make fundamental changes to the ways of the usage of internet and make web sites place for business. Web, connects organizations to their customers and shares a direct media with low costs for services of the organizations or businesses. This method changes the conventional ways of trade and becomes more and more popular in this field. Moreover, for organizations, evaluation of their web-based services and the way of personalizing these services and definition of their competitive advantages based on users' behavior analysis are vital. Hence, discovering hidden knowledge in users' interactions with web attracts more attentions. Organizations and companies need to record, study and analyze their users' behaviors and interests in order to adapt content and interface of their web site to users' interests. Web personalization can be defined as any set of actions that adapts information and services of the web to users' needs based on their characteristics and interests and makes dynamic suggestions according to users' behavioral patterns. For analyzing users' behaviors and making good recommendations, web mining techniques can be used. In this thesis, a model was represented that can be used for analyzing and predicting users' behaviors of a specific web site. First, users were clustered with affinity propagation algorithm and then, their behaviors were analyzed using sequential pattern mining algorithm called CM-SPADE. In the next step, for each cluster, Users' profile was created (users' profile consists of popular web pages among users in a specific cluster). Then by using these profiles, recommendations can be made for new users. At last, the represented model was evaluated and the final results were acceptable.