Bhawna Nigam¹ Sanjiv Tokekar¹ Suresh Jain^2
1Institute of Engineering & Technology, India and ²Institute of Engineering Management & Research, India

Prediction of web user behavior is the demand of today competitive edge of World Wide Web. Predicting the next web page is not sufficient, evaluation of prediction models is important because every model have its own pros and cons. Prediction results will be helpful if high prediction accuracy is achieved with minimum complexity, which are depended on the prediction model. Various models and their variations are proposed for predicting the next web page accessed by the web user. Markov model and their variations are found suitable for web prediction. In this research we have evaluated and compared various models for predicting next web page accessed by the web user. Experiments are conducted on three different real datasets.

Web usage mining, Markov Model, User Navigation Session, web log and web prediction