Diversity versus anonymity for privacy preservation

Abstract

Although k-anonymity prevents disclosure individuals' identity but it to prevent inferring, sensitive information which is aimed by 1-diversity. Most of the recent efforts that address diversity have focused on extending of k-anonymization methods to satisfy diversity as well. In this paper we show that diversity is lonely sufficient to protect private information of individuals and no need to apply k-anonymity first. Moreover I-diversity is stronger than k-anonymity and even some simple proposed techniques (like Anatomy) that consider only diversity are better than advanced k-anonymization techniques from privacy preservation point of view. We show all the cases by different scenarios and explain how diversity outperforms k-anonymity. Only in the case with some restricted assumptions about external data, some k-anonymization techniques give some protection in addition to I-diversity. We show even in this case the anonymity is related to number of tuples in external data instead of k, which is not so realistic.