Skip to content
Outrich Data

Glossary

k-anonymity

k-anonymity is a measure of re-identification risk. A dataset is k-anonymous when every combination of its quasi-identifier values is shared by at least k records. A person in the data cannot then be narrowed to fewer than k candidates by those fields. A k of one means some record is unique.

To measure k, choose the columns that describe people and could be known from outside the dataset, such as job title, location and year of birth. Group the records by the values in those columns. The size of the smallest group is k. A record alone in its group can be matched to a person by anyone who knows those facts about them.

Raising k means making groups larger. Values can be generalized, so a town becomes a region and an exact age becomes a band. Rare records can be removed. Each step loses detail, so the work is a trade between privacy and usefulness. k-anonymity has known limits. It depends on which columns are chosen as quasi-identifiers. It says nothing about free text. And if everyone in a group shares the same sensitive value, that value is revealed even though no individual is singled out, a weakness that l-diversity addresses. No law sets a universal value of k. It is a measure to inform judgment.

The free checker on this site computes group sizes for chosen columns in the browser. It is a quick way to find unique records before deciding which fields to generalize or remove.

Example

In a staff dataset, only one record shows a marine engineer in a given small office hired in a given year. That record has a k of one until the office is generalized to a region.

Find out if your data qualifies.

Eleven questions, about five minutes, no call required. You get an answer on screen and a written follow-up.

esc