Skip to Main Content
Mining generalized association rules is closely related to the taxonomy(is-a hierarchy) data which exists widely in retail, geography, biology and financial domains. If we use traditional method to mine the generalized association rules, it becomes inefficient because the itemsets will be huge along with the items and levels of taxonomy increasing, and it also wastes lots of time to calculate the support of redundant or unnecessary itemsets. In this paper, we proposes a new efficient method called CBP to partition the transaction database into several smaller ones level by level using correlation of itemsets, which make the mining more efficient by reducing the scanning size of transaction database. By experiments on the real-life transaction database, the results show that our CBP_based algorithms outperform the well-known algorithms.