A Fuzzy Markov Model Approach for Predicting User Navigation
Ali A. Ghorbani, Xiaowen Xu · 2007
User navigation is an interesting aspect in Web usage mining. Analysis of this issue can be of great benefit in discovering users' behavior. This paper presents a fuzzy approach for predicting users' navigation paths using the Markov chain model. A standard Markov model can be used to predict the ID of the next page. However, our proposed approach can predict not only users' next requests for pages, but also the time-duration to be spent on the requests. The experimental results show that our method is highly accurate (average 77.9%) in session prediction. Even though the standard methods also perform well (average 78.9%), our proposed approach