K-Anonymity

K-anonymity is a property of an anonymized dataset intended to protect the privacy of the individuals it describes, introduced by Pierangela Samarati and Latanya Sweeney in 1998 and building on an underlying idea from Tore Dalenius 1986 work. It addresses the problem of releasing person specific, field structured data in a way that gives some assurance the individuals described cannot be re-identified, while keeping the data practically useful for research. Formally, a dataset satisfies k-anonymity when the information for each person in the release cannot be distinguished from that of at least k minus one other individuals whose information also appears in the release, meaning every record shares its combination of quasi identifying attributes with at least k minus one others. The concept is applied especially in healthcare, census and research data sharing, though the source notes that the guarantees k-anonymity provides are aspirational rather than strictly mathematical.