Web profiling and navigation path analysis

Abstract

The rich information in the World Wide Web has created an information overloading problem for web users. The technology of Web personalization targets to solve such a problem and thus recommendation systems were studied. Nowadays, Web Usage Mining technique is overtaking the position of traditional collaborative filtering technology, moving traditional ranking style into web log and navigation path analysis approach. The Markov model is one of the well performed prediction model that is suitable for a recommendation system. However, it suffers from state complexity problem and needs to be improved. Association Rule mining is another technique for data mining and pattern discovery. There is also room to improve for the performance of Association Rule. This project focuses on the studies of different optimization schemes for Markov model and Association Rule. It takes experiments on different kinds of improvement studies. It also shows the result of different kinds of model optimization scheme. A profiling modeler and an online recommendation engine are also implemented for testing and simulation.