Privacy-Preserving Publishing Data with Full Functional Dependencies

We study the privacy threat by publishing data that contains full functional dependencies (FFDs). We show that the cross-attribute correlations by FFDs can bring potential vulnerability to privacy. Unfortunately, none of the existing anonymization principles can effectively prevent against the FFD-based privacy attack. In this paper, we formalize the FFD-based privacy attack, define the privacy model (d, l)-inference to combat the FFD-based attack, and design robust anonymization algorithm that achieves (d, l)-inference. The efficiency and effectiveness of our approach are demonstrated by the empirical study.