What is De-identification/Pseudonymization?
De-identification and pseudonymization, told as a story about swapping a guest's name tag for a number, with the real name locked in a separate vault.
The guest list has real names written out in full.Hand the list straight to a research team, and even the names show.
Erasing just the name doesn't stop re-identification.Combine other clues — birthdate, neighborhood — and the guest can be identified again. That's re-identification.
Pseudonymization swaps in a number tag and locks the real name in a separate vault.Only the vault's own key reconnects a number tag to its real name.
Original, pseudonymized, and anonymized are not the same thing.Pseudonymized data is still personal data, since the vault key can reconnect it. Anonymized data has no key table and no clue left to identify anyone, so it isn't personal data.
The more data combined, the higher the re-identification risk.So re-identification risk gets assessed up front. Ledgers from several castles are sometimes combined in a controlled place like the special reading room.
Pseudonymization = swapping a name for a number tag and locking the key table away separately. Anonymization burns that key table and strips the remaining clues, so it's irreversible.
Pseudonymization processes data so an individual can't be identified without additional information; as long as that extra information exists, the data is still legally personal data. Anonymization makes the person unidentifiable even with other information, considering time, cost, and technology reasonably, so it is no longer personal data. Measures like k-anonymity and l-diversity are used to assess whether de-identification is adequate.
When grown-ups say it
- Pseudonymization
- Swapping in a number tag. Reversible with the key table (additional information). Still counted as personal data.
- Anonymization
- Burning the key table and stripping the clues. Re-identification isn't possible, considering time, cost, and technology reasonably. No longer personal data.
- Additional information
- The table linking number to name. The core of pseudonymization. The key table is kept apart from the list, encrypted and locked.
- Re-identification
- Finding the name again from other clues. A combination like birthdate and neighborhood can narrow it to one person.
- k-anonymity
- Grouping k people together. Groups at least k similar guests so no individual stands out.
- Korea's Three Data Acts
- The kingdom's rules for handling data. Korea's Personal Information Protection Act and related laws set the grounds and limits for handling pseudonymized data.
- Data masking
- Hiding only part of it. Similar to pseudonymization, but usually irreversible. → a number instead of a name
- Privacy principles
- The name tag belongs to the guest. Pseudonymization still operates under this principle. → the name tag belongs to the guest