A Multi-Dimensional K-Anonymity Model for Hierarchical Data

Abstract

For improving the usability of the anonymous result, it is important to comply with the hierarchical structure when generalizing quasi-identifying attributes with hierarchical characteristics. We propose an unrestricted multi-dimensional anonymization model which combines global recoding and local recoding methods. The bottom-up anonymization algorithm with the minimal coverage subgraph constraint and the anonymization metric are proposed. The experiment results justify the effectiveness and scalability of this model.