Optimization of Privacy Preserving Mechanisms in Homogeneous Collaborative Association Rules Mining

Abstract

This article focuses on optimization of cryptographic mechanisms used in association rules multiparty mining algorithms with preserving data privacy. The major part of attention is focused on increasing the performance because the computation expense can be prohibitive when applying to large databases. We introduce how to use a Common Decrypting Key for commutative encryption in Secure Set Union to improve performance. As an example of the above mentioned mechanism application, the article presents a new algorithm of mining association rules on horizontally partitioned data with preserving data privacy-CDKSU (Secure Union with Common Decrypting Key). CDKSU is an application of the Common Decrypting Key for a commutative encryption in a Secure Set Union. This algorithm is compared to the KCS scheme (referenced as HPSU also) since they are both based on FDM. As far as the performance optimization is concerned, the application of Elliptic Curve Cryptography versus Exponential Cryptography is presented as well. We believe that this is the first description of application of the Elliptic Curve Pohlig-Hellman Cipher. The system implementing given algorithms is described and subjected to performance tests. Finally, the results of these tests are presented and analyzed.