A Markov prediction model based on page hierarchical clustering

Markov prediction model is the basis of Web prefetching and personalized recommendation. It can be used to extract connotative Web link hierarchy. But the existence of a large amount of Web objects results in data redundancy and model hugeness. This paper presents an improved method that simplifies the topology structure of Web site and extracted the conceptual link hierarchy which can make the organization clearly and legibly. Firstly, Markov Tree is built. Secondly web site link hierarchy is obtained. Then the in-link and out-link similarity are used to find similar pages in the same level which can be clustered together in order to reduce the dimensionality of the transition matrix. Finally, apply above model to predict. Thus the model can help users find information more effectively and efficiently. Experiments based on two real Web log data demonstrate the efficiency of the proposed method, which can not only have good overall performance and clustering effect but also keep relative higher prediction accuracy and recall.