Mining Cross-Transaction Web Usage Patterns

Web Usage Mining is the application of data mining techniques to large web log databases in order to extract usage patterns. However, most of the previous studies on usage patterns discovery just focus on mining intra-transaction associations, i.e., the associations among items within the same user’s transactions. A cross-transaction association rule describes the association relationships among different users’ transactions. In this paper, the closure property of frequent itemsets, which can determine the complete set of all frequent items exactly and is usually much smaller than the latter, is used to mine cross-transaction association rules from web log databases. We give the basic notion of frequent cross-transaction closed itemsets and prove the related necessary theories. And an efficient algorithm, i.e. MFCCPS(Mining Frequent Cross-Transaction Closed Pageviews Sets), is designed and implemented. At last, an extensive experimental result on two synthetic datasets shows that our approach outperforms previous methods.