Generalization:

A de-identification technique that transforms a value into a more general one, essentially reducing its accuracy to protect privacy, while still retaining valuable information for analytics. It is used in situations where a field or combination of fields might form a unique combination which could be used to single out an individual in a dataset, or used to link to other background information to identify an individual.

For example, a person’s age can be generalized by replacing it with an interval (For example, 33 becomes 30-35). A date in the format day/month/year can be generalized by replacing it with just month/year. A category can be generalized by replacing it with a broader category (for example, junior data scientist becomes technical staff).

Return to glossary