Web User Clustering Using Fuzzy and Learning Automata

چکیده مقاله

In recent years, clustering users from Web logs has become an active area of research in Web Mining. Clustering users’ sessions are useful for obtaining user interests to a Web site, and in turn, build user profiles for advanced Web applications, such as Web caching and prefetching. Web datasets often consist of quantitative, vague and incomplete data. This suggests that fuzzyset theory, is a more useful tool to handle such vague and incomplete data and provide better solutions. In this paper, a learning automata based fuzzy clustering algorithm (LAFC) is proposed. The proposed approach is divided into two steps. In the first step, fuzzy weight are assigned to all the web pages using membership function based on their sessions support count, which is transformed to a fuzzy vector. Each element in fuzzy web user sessions represents visited web page and the number of times a web page accessed during a session. Then we put each fuzzy web user session in the nearest cluster using the learning automata. By doing this, a primitive clustering is performed on the web user sessions and the primitive centers of clusters are determined. In the second step, these primitive clusters which have no or several web user sessions are used by weighted fuzzy c-means clustering algorithm and on the basis of the weight of each cluster and the centers of clusters are reclustered. By doing this, the final clusters are determined. Our results show that quality of clusters formed using LAFC algorithm is much better than FCM, Weighted FCM and FNCN.

دریافت لینک دانلود مقاله