A novel approach for horizontal privacy preserving data mining

Abstract

Many business applications use data mining techniques. Small organizations collaborate with each other to develop few applications to run their business smoothly in competitive world. While developing an application the organization wants to share data among themselves. So, it leads to the privacy issues of the individual customers, like personal information. This paper proposes a method which combines Walsh Hadamard Transformation (WHT) and existing data perturbation techniques to ensure privacy preservation for business applications. The proposed technique transforms original data into a new domain that achieves privacy related issues of individual customers of an organization. Experiments were conducted on two real data sets. From the observations it is concluded that the proposed technique gives acceptable accuracy with K-Nearest Neighbour (K-NN) classifier. Finally, the calculation of data distortion measures were done.