Parallel approach in R-Diffset algorithm for infrequent itemset mining

JULAILY AIDA JUSOH; MUSTAFA MAN; WAN AEZWANI ABU BAKAR; MOHD NORDIN ABDUL RAHMAN; SHARIFAH ZULAIKHA TENGKU HASSAN

Author(s)

JULAILY AIDA JUSOH Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin (UniSZA) Besut Campus, 22200 Besut, Terengganu, Malaysia

MUSTAFA MAN Faculty of Ocean Engineering Technology and Informatics, Universiti Malaysia Terengganu (UMT) 21030 Kuala Nerus, Terengganu, Malaysia

WAN AEZWANI ABU BAKAR Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin (UniSZA) Besut Campus, 22200 Besut, Terengganu, Malaysia

MOHD NORDIN ABDUL RAHMAN Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin (UniSZA) Besut Campus, 22200 Besut, Terengganu, Malaysia

SHARIFAH ZULAIKHA TENGKU HASSAN Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin (UniSZA) Besut Campus, 22200 Besut, Terengganu, Malaysia

Abstract

Data Mining is an established methodology for obtaining critical information from databases utilizing the Association Rule Mining (ARM) technique. This vital information can lead to the association rule, which may reveal a good trend. Association rules' beneficial pattern is frequently stated as frequent and infrequent. To perform Itemset mining, the data formats that are necessary are horizontal format and vertical format. The vertical data format is focused on current research trends in infrequent mining techniques. An example of a vertical data mining technique for an infrequent pattern is called Rare Incremental Equivalence Class Transformation or shorten as R-Eclat. Out of the four variations of the R-Eclat approach, this study will only focus on the R-Diffset form. Prior studies have shown that the R-Diffset algorithm's data processing execution time is time-consuming. The R-Diffset technique is a solution to load imbalance problems that employ numerous nodes and cluster power to provide a novel parallel approach. In response to the positive findings of mining in terms of faster processing time and less memory consumption, R-Diffset will be supplemented with a parallel technique. Finally, a novel parallel strategy is provided to overcome the restrictions of sequential processing in terms of speed.

Publication Details

Page(s) 4761-4770

DOI DOI not available

Published Journal: Journal of Theoretical and Applied Information Technology, Volume: 100, Issue: 24, Year: 2022

Request the full text from PASTIC

Keywords

Association rule mining (ARM), Data Mining, Infrequent mining, R-Diffset, Rare Incremental Equivalence Class Transformation (R-Eclat)