Mining multiple-level association rules from a transactional database using FP-tree

In many database applications, information stored in a database has a built-in hierarchy consisting of multiple levels of concepts. In such a database, users may want to find association rules among items only at the same level or association rules that span over different levels. This task is called multiple-level association rule mining. An Apriori-based algorithm was proposed to do this task [3, 4]. But, it has to scan the database as many times as determined by the length of the longest frequent itemset. In this paper we introduce a new algorithm, called MFP-tree, which is based on the FP-tree algorithm [5]. The proposed algorithm requires only two scans of the database and our performance stud), shows the MFP-tree algorithm outperforms the Apriori-based algorithm for mining multiple-level association rules.