deanonymizable

Deanonymizable refers to the degree to which data that has been anonymized can still be linked to the individuals to whom it pertains, thereby compromising or removing anonymity. In privacy literature, deanonymization describes both the potential and the process by which re-identification occurs, often through the combination of anonymized data with external information.

Factors that influence deanonymizability include the presence of quasi-identifiers—attributes that are not unique on their own but become identifying when combined—data granularity and sparsity, the overlap between released data and public records, and the availability of auxiliary datasets. Even if direct identifiers such as names or numbers are removed, combinations of attributes like birth year, postal code, and gender can narrow down or reveal identities when linked with other sources.

Contexts where deanonymizability is a concern include health data releases, mobility and location datasets, consumer transaction records, and social media analytics. The risk highlights the tension between data utility for research or service improvement and privacy protection for individuals.

Mitigation strategies aim to reduce deanonymization risk, including data minimization, access controls, and privacy-preserving techniques such as differential privacy or data perturbation. Organizations may also implement governance frameworks, auditing, and legal safeguards to manage re-identification risk and to ensure compliance with privacy laws.

See also de-anonymization, anonymization, privacy, differential privacy.