MINING CONCISE REPRESENTATIONS OF FREQUENT HIGH-UTILITY OCCUPANCY ITEMSETS USING GENERATOR PATTERNS

A current trend in data mining is the discovery of frequent high-utility occupancy itemsets (FHUOIs) in quantitative databases. These itemsets capture user preferences and significantly contribute to transaction utility, making them valuable for real-world applications. However, analyzing the result set of all FHUOIs is complicated by its often-massive size, especially in large or dense datasets with low minimum-utility occupancy thresholds. To address this challenge, this paper introduces GFHUOI-Miner, a novel algorithm designed to efficiently extract a smaller set of all generators of FHUOIs (GFHUOIs). The algorithm reduces the search space by early elimination of non-generator FHUOI branches using a novel weak lower bound on the utility occupancy function, named pFEwlb, along with two efficient pruning strategies based on pFEwlb. Additionally, GFHUOI-Miner incorporates an optimization technique to minimize computationally expensive inclusive relationship checks among itemsets. Experimental results demonstrate that the proposed algorithm outperforms the baseline in terms of runtime, memory usage, and scalability for discovering all GFHUOIs.