期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：10
出版社：Journal of Theoretical and Applied
摘要：Association rule mining is the process of identifying the frequent items and associative rules in a market basket data analysis for large set of transactional databases. Association rules are employed in different data mining applications including web mining, intrusion detection and bioinformatics. In recent years it has been seen tremendous advances in the ability to perform effective association rule mining. It causes the need of sensitive rule selection to enhance the privacy preservation of data transactions. A new technique called Gravitational Search Based Sensitive Rule Selection (GS-SRS) is proposed to select the sensitive rules is to be hided for improving the privacy preservation of transactional database. The GS-SRS technique is introduced to select the sensitive rules from the derived association rule through conditional probability. The sensitive rules contain the sensitive information of transactional database. Sensitive rules are identified for many applications. One of the applications of sensitive rule identification is to preserve the privacy of an organization or an individual by hiding these rules. The GS-SRS technique initially generates association rules through identifying the frequent items by using conditional probability-based association rules and support count and confidence value. Next, GS-SRS technique used gravitational search algorithm that lists the cohesive and non-cohesive items for the given transactional database in shorter time than the conventional means of feature selection using Rough Set technique. The association rule containing more cohesive items is selected as sensitive rule. A threshold value is used to select sensitive rules with the convergence of cohesive items and divergence of the non-cohesive items. Finally, the sensitive rules selected are hided for preserving the privacy of transactional database. The experiments have been carried out on transaction database using four data sets and compared with state of art existing techniques. The experiment results show that the proposed GS-SRS technique is able to improve the accuracy of privacy preservation with minimum execution time when compared to state-of-the-art works.
