Anonymisation is the process of transforming data so that individuals can no longer be identified, directly or by inference, while preserving enough utility for analysis. Techniques include suppression, generalization, pseudonymisation, k-anonymity, differential privacy, and the blurring or synthetic replacement of faces and identifiers in media. Effective anonymisation must resist re-identification through linkage with auxiliary data.
Content
- Methods trade privacy against utility: aggregation and generalization reduce granularity, differential privacy adds calibrated noise with formal guarantees, and synthetic substitution replaces real identifiers with realistic fakes. The persistent risk is re-identification by linking quasi-identifiers to external datasets, so strong anonymisation is evaluated against realistic attacker models rather than assumed from naive masking.