There have been proposed many algorithms for frequent pattern mining. The first efficient algorithm was proposed by Aggarwal et al. which is called Apriori [10]. The most weaknesses of this algorithm are multi scan of database and generating many candidate patterns. To solve these weaknesses, J. Han et al. proposed FP-growth [11] in 2000. It aims to enhance the efficiency of frequent pattern mining by using a novel tree structure called FP-tree. Efficiently, data is captured and kept by FP-tree by two database scans. Then, to mine frequent patterns, FP-growth algorithm explores the tree in a divide-and-conquer strategy without candidate pattern generation. Although the experimental results show that the performance of FP-growth is very higher than Apriori, FP-growth is not adoptable to interactive mining of frequent patterns because FP-tree kept only frequent patterns.