Chapter 3: Introduction to Technological Aspects of Privacy
Approaches to Deidentification: Suppression, Generalization, Noise Addition
Three core techniques hide identity: suppression removes values, generalization replaces detail with a broader category, and noise addition substitutes similar-but-different values while preserving statistics.
| Technique | How it works | Example |
|---|---|---|
| Suppression | Remove identifying values from a record | Dropping names/phone numbers before statistical analysis |
| Generalization | Replace detail with a broader category | Year of birth (~80 categories) instead of full date (~25,000); municipality instead of GPS |
| Noise addition | Replace values with similar but different ones, preserving statistics | Altering precise salaries so payroll cannot match an individual, while keeping the average |
Key terms - quick answers
What is “Suppression”?
Removing identifying values from a record (e.g. dropping customer names for statistical analysis).
What is “Generalization”?
Replacing a detailed data element with a more general one (e.g. year of birth instead of full date, municipality instead of GPS).
What is “Noise addition”?
Replacing actual values with similar but different ones, often preserving statistical properties like the average while disrupting individual identification.