Marcov Models for Web Access Prediction

The problem of predicting user’s behavior on a Web site has fundamental significance due to the rapid growth of the World Wide Web. Although traditional Markov models have been found to be suited for addressing this problem, they have serious limitations. Thus, good predictions require new Markov models. Hybrid-order tree-like Markov models predict Web access precisely while providing high coverage and scalability. Selective Markov models intelligently select parts of different order Markov models such that the resulting model has a reduced state of complexity, while maintaining a high predictive accuracy. The goal of our work is to present and compare traditional Markov models and new Markov models, regarding the prediction of users behavior.