Approaches for preserving FDs in k-anonymization
Jinling Song, Guangbin Zhang, Liming Huang, Xingshun Liu, Danli Wang · 2010
K-anonymization essentially is some update operations over the original dataset. So, to guarantee the integrity of the dataset, it's necessary to preserve the functional dependencies (FDs) in k-anonymization. We present several approaches to maintain FDs in k-anonymization. One is detecting FDs violation constantly while k-anonymizing, which can be merged to numerous previous k-anonymized algorithms. Another is based on clusters combination, which is suit for k-anonymized algorithms using clustering or microaggregation. The third is a more directly and valid approach based on K-MSD and associated generalization, which focuses on preserving FDs as well as higher data precision and increases the utility of the anonymized dataset effectively.