Characterizations of Local Recoding Method on k-Anonymity

Abstract

k-Anonymity is one of the most widely used techniques for protecting the privacy of the publishing datasets by making each individual not distinguished from at least k-1 other individuals. The local recoding method is an approach to achieve k-anonymization through suppression and generalization. The method generalizes the dataset at the cell level. Therefore, the local recoding could achieve the k-anonymization with only a small distortion. As the optimal k-anonymity has been proved as the NP-hard problem, the plenty of optimal algorithm local recoding has been proposed. In this research, we study the characteristics of the local recoding method. In addition, we discover the special characteristic dataset that all generalization hierarchies of each quasi-identifier are identical, called an “Identical Generalization Hierarchy” (IGH) data. We also compare the efficiency of the well-known algorithms of the local recoding method on both \(non-IGH\) and IGH data.