A Novel Approach of Association Rule Hiding for Preserving Privacy in Data Mining

Abstract

Organizations use customer relationship management for making a healthy relationship with
their existing customers. But, there are many privacy issues in information collection,
utilization and sharing. Information breach is always a concern for customers as they are
always afraid of losing their private information to others. On the web, a person has to
provide his private information along with some rights if he/she creates any kind of account
or channel. Afterwards, incessant email or promotional publications, even telephone calls,
will bother clients. More seriously, firms may sometimes know the number and location of
members in the family and even their school in childhood. All these factors put the client at
risk. Many clients are troubled by the scenario. This has negative consequences for excellent
customer relationship management. In the meanwhile, the rate of information abuse has also
increased. An algorithm named RDALR, which enhances the privacy of sensitive knowledge,
is proposed in this paper, namely the Rectified Deletion and Addition in LHS of Rule. The
proposed algorithm provides better efficiency in comparison to existing algorithm. Both GPU
and non-GPU runtime are provided. They are both quicker than existing algorithms.
Experiments demonstrate the effectiveness and privacy requirements of the proposed
method. In addition, it ensures reliable results of data mining process with preservation of
confidential information.