Preserving FDs in K-Anonymization by K-MSDs and Association Generalization
Jinling Song, Liming Huang, Qi He, Yan Gao, Xin Liu, Yuxiang Li · 2009
Although k-anonymity can guarantee the security of privacy, it may violate data dependencies, such as FDs (functional dependency) in k-anonymization. We define a new data dependency named k-multiset dependency (K-MSD), and show that a K-MSD dataset satisfies k-anonymity constraint too. So, it is possible to implement k-anonymization through constructing K-MSDs over original dataset. For the FDs over the original dataset, we preserve them using association generalization (AG) while constructing K-MSDs. Then, we propose a k-anonymization algorithm: K-MSD-AG to preserve FDs.