Discovering web access orders with association rules

In the past few years, the use of World-Wide-Web (WWW) has grown exponentially. It is important for companies to analyze the behaviors of their customers so as to have better profit and services. This paper explores the technique of association rules to discover the access patterns of WWW users. In order to support our proposed algorithms, there is a pre-mining phase which filters and transforms WWW access logs into a database of access transactions. Our first algorithm, Extended Apriori algorithm (EAA), is a variant of the infamous Apriori algorithm. This algorithm is modified to handle the item order in the counting and generation steps of the candidate access patterns. However, the EAA does not deal with the minimal viewing time constraint directly. In the OPM algorithm, we propose to consider how to reduce the number of database scans and exploit the minimal viewing time constraint during the candidate itemset generation. In our experiments, the results showed that the OPM, algorithm took less than half of the time required by by the EAA