An New Attractive Mage Technique Using L-Diversity
Ganesan Anandhi, Kumarasamy Saravanan · Machine Learning and Applications An International Journal · 2016
Data that is published or shared between organizations contain private information about an individual. The concept of Privacy Preservation aims to preserve this sensitive information from various privacy threats that violate the privacy of an individual. Analysis of this private information could revealinformation that can be used for malicious purposes by the attackers.Anonymization is a privacy preservation approach suitable for mixed data that contains both numerical and categorical attributes.In this paper a novel method called Micro-aggregation Generalization (MAGE) is used for anonymization of microdata that can retain more semantics of the original data.Here the Micro-aggregation is applied over the numerical data and Generalization is applied over the categorical data.Even though the MAGE approach preserves privacy it fails to address the homogeneity and background knowledge attacks.Later the l-diversity approach is applied to deal with homogeneity attack.In l-diversity, the anonymized records are reordered to satisfy a new privacy principle that removes homogeneity of sensitive information.The result shows that the MAGE approach suffers from homogeneity attack and applying l-diversity over MAGE prevents homogeneity attack and also provides better privacy and data utility.