A Privacy Attack on Multiple Dynamic Match-key based Privacy-Preserving Record Linkage.

Area of Science:

• Computer Science
• Information Security
• Data Privacy

Background:

• Increasing demand for cross-database record linkage necessitates privacy-preserving techniques.
• Multiple dynamic match-key encoding offers privacy but risks re-identification via frequency analysis in large datasets.

Purpose of the Study:

• To introduce a frequency-based attack to assess the privacy of multiple dynamic match-key encoding.
• To propose actionable recommendations to mitigate identified privacy vulnerabilities.

Main Methods:

• Developed a frequency-based attack analyzing match-key distributions against a known plain-text database.
• Utilized statistical correlation tests and frequency alignment for re-identification.
• Proposed two methods to modify frequency distributions to prevent attacks.

Main Results:

• The attack successfully re-identified sensitive values encoded with multiple dynamic match-key encoding in real-world databases.
• Achieved up to 98% precision and recall in re-identifying plain-text values.
• Proposed recommendations effectively hindered the attack with minimal impact on linkage quality.

Conclusions:

• The frequency-based attack highlights critical privacy weaknesses in current match-key encoding methods.
• The proposed recommendations enable secure application of multiple dynamic match-key encoding for sensitive data linkage.