Tree-Based Unified Temporal Erasable-Itemset Mining

Erasable itemset mining is an important research area for manufacturers, as it aids in identifying less profitablematerials in product datasets to facilitate better decision-making for managers. It allows for improved trade-offs between manufacturing and procuring activities. Traditional erasable itemset mining does not account for the time factor, which is critical for time-sensitive industries, with product time periods significantly impacting a company's profitability. Therefore, we previously proposed an Apriori-based unified temporal erasable itemset mining approach, which could consider different user scenarios to solve this issue. In this work, we design a tree-based algorithm to raise the mining efficiency. It applies a lower-bound strategy to reduce the candidate erasable itemsets and the number of database scanning. According to the experimental results, our proposed algorithm has better performance on execution time and memory usage than the previous work.