👥
k‑Anonymity
Anonymization
Privacy model where each record is indistinguishable from at least k−1 others on quasi-identifiers.
Definition
k‑anonymity requires that for chosen quasi-identifiers (e.g., age, ZIP, gender), each equivalence class has size ≥ k. Achieved via generalization and suppression, but does not fully prevent attribute inference.
In plain English
Privacy model where each record is indistinguishable from at least k−1 others on quasi-identifiers.
Why this matters
Why it matters: It provides baseline re-identification resistance but can be vulnerable to linkage or homogeneity attacks.
Example
Example: Generalize ZIP codes and ages until each group has at least k records before releasing the dataset.