I. Overview

%%{init: { 'theme': 'base', 'themeVariables': { 'edgeLabelBackground': '#fff' }}}%%
flowchart LR
    A["Identifiable\ninformation"] -- "applying the 5 core\npseudonymization techniques" --> B["Pseudonymized\n(de-identified) data"]
    style A fill:#f9f9f9,stroke:#333,stroke-width:3px
    style B fill:#e1f5fe,stroke:#01579b,stroke-width:3px

Definition: Pseudonymization techniques are technical methods that delete or substitute part of personal information so that a specific individual cannot be identified without additional information.

Core value and purpose:
( Data Utility ) Preserves the data’s statistical properties and quality ( Utility ) so it still fits the intended analysis purpose.
( Privacy Protection ) Minimizes re-identification risk to safeguard data subjects’ privacy rights.
( Risk Management ) Secures legal compliance through technical measures and cuts off the risk of a security incident (re-identification) at the source.

II. Mechanism & Components

The 5 Core Pseudonymization Techniques and Their Sub-Methods

flowchart LR
    A["Original\npersonal information"] --> T1["1. Deletion"]
    A --> T2["2. Rounding"]
    A --> T3["3. Masking"]
    A --> T4["4. Substitution"]
    A --> T5["5. Aggregation"]

    T1 --> B["Pseudonymized data\nbalancing Utility\nand Privacy"]
    T2 --> B
    T3 --> B
    T4 --> B
    T5 --> B
Pseudonymization TechniqueDetailed DescriptionConcrete Example
1. DeletionDirectly deletes fields containing identifiers, or removes specific itemsFully deleting name, resident registration number, phone number
2. RoundingConverts numerical data into a range (bucket) or rounds itAge 26 → “20s”; salary 3.54 million won → “3-4 million won”
3. MaskingReplaces part of the data with a special character such as an asterisk (*) to prevent identificationHong Gil-dong → Hong *dong; Gangnam-gu, Seoul → **-gu, Seoul
4. SubstitutionReplaces an identifier with an arbitrary unique or virtual numberResident registration number → replaced with a serial number (A-001)
5. AggregationProcesses data as a group total or average rather than individual recordsIndividual income → converted to average income by department

III. Advanced Topics & Comparison

Pseudonymization techniques are more often mixed according to the characteristics of the data than used alone, and the following privacy-protection models are compared when applying them.

Comparison Itemk-Anonymityl-Diversityt-Closeness
Core ConceptKeeps at least k records sharing the same attributesRequires at least l distinct sensitive values within each k-groupMinimizes the difference between the overall attribute distribution and a specific group’s distribution
Threat DefendedLinking attackHomogeneity attack, background-knowledge attackSkewness attack, similarity attack
LimitationVulnerable to homogeneity attacksMay reduce data utilityComplex and costly to implement

Last updated 18 Aug 2026, 00:00 UTC. history