A Lattice-based approach for Mining Non-Redundant High Utility Association Rules

High utility itemsets mining provides more information, which can help manager to make better decisions. However, there are too many redundant high utility itemsets which makes it difficult for users to find useful information in a large set of high utility itemsets. To solve this problem, this paper proposed an approach for mining non-redundant high-utility association rules (NR-HAR) efficiently in both dense and sparse datasets, which make more timely decisions on business strategy. This approach includes two phases: (1) building an extended high utility itemsets Lattice(EHUIL) and identify generator and closed itemsets within this. (2) Proposing an efficient algorithm for mining NR-HAR from the built EHUIL. The new algorithm is based on utility-confidence framework and the lattice structure, obtaining semantic relationships among HUIs. Extracting results from EHUIL, which having closed flag and generator flag, determining to find rules with child nodes to be stopped or continued. We carry out experiments to show the efficiency of the proposed method. The experimental results show that this new approach has a faster runtime on different types of datasets than existing methods, especially in sparse database.