A Comparison of anonymization protection principles

摘要

We do a survey of some of the most important principles of anonymization present in the literature. We identify different kinds of attacks that can be thrown against an anonymized dataset and give formulas for the maximum probability of success for each. For each principle, we identify whether it is monotonous, what attacks it is suited to counter, if any, and what principles imply other principles. We end by giving a classification of anonymization principles and giving guidelines to choosing the right principle for an application. Although we could not cover all principles in the literature, this is a first step to a systematization and simplification of proposals for anonymization principles.