Privacy Preserving Data Utility Mining Using Perturbation

Abstract

Data Mining is a field of research dealing with the automatic discovery of knowledge within databases. Recent advances in data mining has increased the disclosure risks that one may encounter when releasing data to outside parties. Privacy preserving data mining (PPDM) deals with protecting the privacy of individual data or sensitive knowledge without sacrificing the utility of the data. Privacy Preserving Utility Mining (PPUM) is an extension of PPDM where the quantity as well as the utility are taken care of. Perturbation is a technique which modifies the contents of database with constraints and satisfies the privacy policies of the data holder. A Fast Perturbation using Frequency Count (FPUFC) algorithm is proposed to hide all sensitive high utility itemsets. The performance of proposed algorithm were compared with that of the existing algorithm, Fast Perturbation using Tree and Table Structures (FPUTT). FPUFC shows better performance by taking lesser execution time compared to FPUTT.