Data Privacy Preservation Algorithm on Large-Scale Identical Generalization Hierarchy Data

Abstract

In the data bursting era, an enormous amount of data can be collected. Though, such data can benefit both the data owners and the public, the privacy of collected data is an important concern. To guarantee the privacy of data, the k-anonymous method is applied before publishing the dataset. The dataset needs to have an identical value of at least k records in order to protect the data privacy which could cause data losses. The optimal k-anonymity is concerned with minimizing the data losses and also preserve data privacy. The generalization lattice is created to map all the generalization schemes and use them to find the optimal answer. The larger number of attributes means the larger number of nodes in the generalization lattice. Thus, in the larger number of attributes, the k-anonymous algorithm takes more computation resources and time to determine the answer. Although, due to the limited computation resources and time, the existing optimal k-anonymity algorithms only find the optimal answer on the small dataset. Therefore, in this paper, we design the optimal k-anonymity algorithm with the incremental attribute concept. At m attribute, our algorithm process only the nodes which previously satisfy the k-anonymity from the $m-1$ attributes. Thus, our algorithm can find the optimal answer at a large number of attributes without reaching the memory limit problem compared with the existing k-anonymity algorithms due to it determines only some necessary nodes in the lattice.