Generalization-Based k-Anonymization

Microaggregation is an anonymization technique consisting on partitioning the data into clusters no smaller than k elements and then replacing the whole cluster by its prototypical representant. Most of microaggregation techniques work on numerical attributes. However, many data sets are described by heterogeneous types of data, i.e., numerical and categorical attributes. In this paper we propose a new microaggregation method for achieving a compliant k-anonymous masked file for categorical microdata based on generalization. The goal is to build a generalized description satisfied by at least k domain objects and to replace these domain objects by the description. The way to construct that generalization is similar that the one used in growing decision trees. Records that cannot be generalized satisfactorily are discarded, therefore some information is lost. In the experiments we performed we prove that the new approach gives good results.