Clustering-Based Categorical Data Protection

The need of improving the privacy on public datasets is be- coming more and more important because the number of public available datasets is growing very fast. This forced the continuous research to find better protection methods that prevent the disclosure of the entities or individuals in a dataset while preserving the data utility. In this paper we present a new approach for categorical data protection based on applying clustering to the dataset and then protecting each cluster. We show that this new approach allow us to have protections with better trade-off between data utility and individuals information disclosure.