Statistical approaches to de-identification

In order to protect datasets that contain personal or otherwise sensitive data, there is an increasing number of statistical approaches to de-identification, which to some extent quantify how identifiable data are after de-identification.

In this chapter, we discuss the following approaches, as these are the most widely used approaches:

• K-anonymity
• L-diversity
• T-closeness
• Differential privacy

These approaches (or: privacy models) are not yet much used in research practice, because they come with some disadvantages and require resources and/or expertise to be applied and interpreted correctly. However, they are used in many de-identification tools and are useful to detect specific sensitivities in (tabular) datasets. For those reasons, the techniques are introduced in this chapter.

---

Data Privacy Handbook

The information presented here is provided as is, with no guarantees of accuracy or completeness. For the most up-to-date information, please refer to your privacy officer, the university website or intranet. We cannot be held responsible for any negative consequences due to incorrect interpretation or use, and inconsistencies with policies/views of other institutions.

Privacy policy | Cite the Data Privacy Handbook