Using Distortion and Blocking Techniques for Preventing Association Rules' Discovery

Abstract

Background: Data mining and privacy preserving are scientific concepts, the former of which detects the relationships among items of databases and the latter preserves sensitive rules. Their disclosure will threaten the database owner's security. Methods: Sanitizing the database by privacy preserving algorithms causes some side effects such as lost rules, dissimilarity, etc. All privacy preserving algorithms attempt to reduce these side effects. In addition to the side effects, inferring sensitive data from insensitive data should also be taken into account. Results: In this paper, a new technique is presented in order to block inference channels and reduce lost rules caused by database sanitization. In this paper, combination of distortion and blocking techniques enabled the researcher to block the inference channels and also to reduce lost rules and CPU usage compared to CR algorithm. Finally, the proposed algorithm was assessed with CR and DSR algorithms and the assessment results indicated efficiency of it. Applications: The proposed algorithm had better performance compared to the CR. It was attempted to proximate the results obtained by the suggested algorithm to those obtained by the DSA.