Novel concise representations of high utility itemsets using generator patterns

Abstract

Mining High Utility Itemsets (HUIs) is an important task with many applications. However, the set of HUIs can be very large, which makes HUI mining algorithms suffer from long execution times and huge memory consumption. To address this issue, concise representations of HUIs have been proposed. However, no concise representation of HUIs has been proposed based on the concept of generator despite that it provides several benefits in many applications. In this paper, we incorporate the concept of generator into HUI mining and devise two new concise representations of HUIs, called High Utility Generators (HUGs) and Generator of High Utility Itemsets (GHUIs). Two efficient algorithms named HUG-Miner and GHUI-Miner are proposed to respectively mine these representations. Experiments on both real and synthetic datasets show that proposed algorithms are very efficient and that these representations are up to 36 times smaller than the set of all HUIs.