Predicting User Navigation in an Online Newspaper Site Using Association Rules Mining and Markov Model

Abstract

This paper discusses an approach to predict Web pages from an online newspaper using association rules mining and Markov model decision process. We use a set of Web server logs from an online newspaper, process the logs using Web usage mining methodology, generate transaction files for association mining and predict the web pages using Markov decision model process. We found that users are reading articles from the same section and since majority of users only read one page in a session, it is hard to find associated news articles in a same session. However, the association between section pages are legit and can be used to model the Markov chain for the navigation.