Checking anonymity levels for anonymized data

Abstract

Privacy Preserving Publication has become one of the most prominent research topics in the recent years. Several techniques like k-anonymity, l-diversity and (α, k) anonymity were proposed to preserve privacy. Most of the published work focuses on anonymizing the microdata for preserving privacy and now the focus towards the verification of the anonymity levels of the microdata before publishing is the need of the day. Many publishers claim having anonymized the data. Verification of the claim on a large anonymized dataset is a herculean task. This paper focuses on providing simple approach for checking the anonymity levels for an anonymized dataset using frequent itemset generation. A GUI based tool named PRUDENT was developed to demonstrate the practicality of the solution. PRUDENT deals with numerical, categorical and multiple sensitive attributes. Results show that the algorithm is feasible and practical. A comparison with the existing methods is shown. © 2011 Springer-Verlag.