Two-Phase Algorithms for a Novel Utility-Frequent Mining Model Jieh-Shan Yeh 1 , Yu-Chiang Li 2 , and Chin-Chen Chang 3,4 1 Department of Computer Science and Information Management, Providence University, Taichung 433, Taiwan jsyeh@pu.edu.tw 2 Department of Computer Science and Information Engineering, Southern Taiwan University, Yung-Kang City, Tainan 71005, Taiwan lyc@cs.ccu.edu.tw 3 Department of Information Engineering and Computer Science, Feng Chia University, Taichung 40724, Taiwan 4 Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi 62102, Taiwan ccc@cs.ccu.edu.tw Abstract. When companies seek for the combination of products which can constantly generate high profit, the association rule mining (ARM) or the utility mining will not achieve such task. ARM mines frequent itemsets without knowing the producing profit. On the other hand, the utility mining seeks high profit items but no guarantee the frequency. In this paper, we propose a novel utility-frequent mining model to identify all itemsets that can generate a user specified utility in transactions, in which the percentage of such transactions in database is not less than a minimum support threshold. A utility-frequent itemset indicates that such combination of products can constantly generate high profit. For finding all utility-frequent itemsets, there is no efficient strategy due to the nonexistence of “downward/upward closure property”. In order to tackle such challenge, we propose a bottom-up two-phase algorithm, BU-UFM, for efficiently mining utility-frequent itemsets. We also in- troduce a novel concept, quasi-utility-frequency, which is upward closed with respect to the lattice of all itemsets. In fact, each utility-frequent itemset is also quasi-utility-frequent. A top-down two-phase algorithm, TD-UFM, for mining utility-frequent itemsets is also presented in the paper. 1 Introduction Data Mining has made a profound impact on business practices and knowl- edge management in recent years. Association Rule Mining (or market basket analysis), finding interesting association or correlation relationships among data items, is one of the most important data mining strategies. Since the concept of association rules was introduced by Agrawal et al. [2] in 1993, many algorithms T. Washio et al. (Eds.): PAKDD 2007, LNAI 4819, pp. 433–444, 2007. c  Springer-Verlag Berlin Heidelberg 2007