Efficient Top-K Identical Frequent Itemsets Mining without Support Threshold Parameter from Transactional Datasets...

Area of Science:

• Data Science
• Artificial Intelligence
• Business Analytics

Background:

• Internet of Things (IoT)-enabled smart shopping carts generate vast amounts of data.
• This data offers potential for deriving business strategies and goals.
• Artificial Intelligence (AI) methods, particularly Association Rule Mining (ARM), are crucial for extracting insights from this big data.

Purpose of the Study:

• To address the computational challenges in top-K frequent itemsets mining (FIM) without user-tuned support thresholds.
• To propose an efficient algorithm for discovering top-K identical frequent patterns.

Main Methods:

• Developed the TKIFIs Miner algorithm utilizing a depth-first search strategy.
• Implemented specialized one- and two-itemsets-based pruning techniques for efficient mining.
• Conducted comparative analysis on benchmark datasets (e.g., Retail, T40I10D100K, T10I4D100K).

Main Results:

• The TKIFIs Miner algorithm demonstrates superior performance in top-K frequent pattern mining.
• It outperforms existing methods that do not rely on user-defined support thresholds.
• The algorithm efficiently handles large datasets and complex itemsets.

Conclusions:

• The proposed TKIFIs Miner algorithm offers an efficient solution for top-K frequent pattern mining from big data generated by smart shopping carts.
• This advancement can enhance business intelligence and strategic decision-making in retail environments.
• The algorithm's pruning techniques contribute to its computational efficiency and effectiveness.