Abstract:
Traditional frequent itemsets mining (FIM) suffers from the vast memory cost, small processing speed and insufficient disk
space requirements. FIM assumes only binary frequency value for items in the dataset and assumes equal importance value for
items. In order to target all these limitations of FIM, high-utility itemsets (HUIs) mining has been presented. HUIs mining is
more complicated and difficult than FIM. HUIs mining algorithms spend more execution time because of large search space.
Therefore, soft computing techniques-based HUIs mining has been proposed. Soft computing techniques provide a systematic
process to discover the optimum solutions by using the concept of natural evolution. This article explores the usage of soft
computing techniques in HUIs mining. The article presents a taxonomy of soft computing techniques-based HUIs mining
including evolutionary computation and fuzzy logic-based approaches. This article enhances understanding of HUIs mining
problems, the current status of provided solutions. The paper provides a comparison analysis of key techniques and discusses
theoretical aspects of the various nature-inspired and fuzzy logic-based techniques along with their pros and cons. In addition,
an overview of strategies and approaches is also presented. Finally, the article mentions key research opportunities and future
directions.
