ABSTRACT
The raw data table usually does not satisfy a specified privacy requirement and the table must be modified before being published. The modification is done by applying a sequence of anonymization operations to the table. An anonymization operation comes in several flavors: generalization, suppression, anatomization, permutation, and perturbation. Generalization and suppression replace values of specific description, typically the QID attributes, with less specific description. Anatomization and permutation de-associate the correlation between QID and sensitive attributes by grouping and shuffling sensitive values in a qid group. Perturbation distorts the data by adding noise, aggregating values, swapping values, or generating synthetic data based on some statistical properties of the original data. Below, we discuss these anonymization operations in detail.
