Explainable Itemset Utility Maximization with Fuzzy Set

Abstract

Recently, fuzzy utility pattern mining has received much attention for its practicality and comprehensibility. It aims to discover high fuzzy utility itemsets (HFUIs) by considering not only utility but also linguistic factors. Among existing algorithms, experiments showed that fuzzy-list-based algorithms are effective and efficient. However, a significant disadvantage of fuzzy-list-based algorithms is that constructing and maintaining fuzzy-lists is time-consuming and memory-overhead. To address this issue, a novel algorithm named explainable Itemset Utility Maximization with Fuzzy Set (FS-IUM) is proposed in this paper. The traditional fuzzy-list structure is replaced by a better structure (i.e., fuzzy-list buffer), which speeds up the mining process and reduces memory consumption. Compared with fuzzy-list structure, the fuzzy-list buffer structure and its auxiliary structure help the algorithm locate the fuzzy-list quickly and thus reduce the runtime. Moreover, with an efficient fuzzy-list buffer construction method, the algorithm reduces the cost of candidate storage. Furthermore, with several efficient strategies, the proposed algorithm can prune numerous useless patterns in advance and thus considerably reduces the runtime usage. Finally, extensive experiments on various datasets were conducted to compare the performance of FS-IUM with some state-of-the-art algorithms. The experimental results reveal that the proposed fuzzy-list buffer-based algorithm highly outperforms the baselines in terms of runtime and memory consumption.