Using pruning and filtering strategies to speed-up projection-based utility mining

In this paper, we try to improve the performance of utility mining. We propose a new projection-based mining algorithm and embed two pruning strategies in it to efficiently find high utility itemsets in a database. By using the two designed strategies, a large number of unpromising itemsets can be pruned away at an early stage, and the data size could recursively be reduced to save the scan time. Finally, the experimental results on synthetic datasets show the proposed algorithm runs faster than the other utility mining algorithms under different parameter settings.