Attribute Based Anonymity for Preserving Privacy

Abstract

Privacy Preserving Publication has become much concern in this decade. Data holders are simply publishing the dataset for mining and survey purpose with less knowledge towards privacy issues. Current research has focused on statistical and hippocratic databases to minimize the re-identification of data. Popular principles like k-anonymity, l-diversity etc., were proposed in literature to achieve privacy. There is a possibility that person specific information may be exposed when the adversary ponders on different combinations of the attributes. In this paper, we analyse this problem and propose a method to publish the finest anonymized dataset that preserves both privacy and utility.