A method of security improvement for privacy preserving association rule mining over vertically partitioned data

Abstract

There have been growing interests in privacy preserving data mining. Secure multiparty computation (SMC) is often used to give a solution. When data is vertically partitioned scalar product is a feasible tool to securely discover frequent itemsets of association rule mining. However, there may be disparity among the securities of different parties. To obtain equal privacy, the security of some parties may be lowered. This paper discusses the disharmony between the securities of two parties. The scalar product of two parties from the point of view of matrix computation is described. We present one algorithm for completely two-party computation of scalar product. Then we give a method of security improvement for both parties.