Efficient Privacy-Aware Record Integration.

Area of Science:

• Computer Science
• Data Privacy
• Database Management

Background:

• Accurate data analysis requires integrating information from multiple repositories.
• Privacy-preserving record linkage (PRL) is essential for matching records across datasets while protecting confidentiality.
• Existing PRL methods like secure multi-party computation (SMC) are computationally expensive, and data transformation techniques risk information leakage and false matches.

Purpose of the Study:

• To introduce a novel, practical model for privacy-preserving record linkage.
• To achieve controlled information leakage and prevent false matches.
• To enhance the efficiency and privacy of record linkage.

Main Methods:

• Data sources are partitioned into blocks to reduce comparisons.
• An efficient secure multi-party computation (SMC) technique is applied to candidate record pairs.
• Differential privacy is used to obfuscate data, providing controlled leakage and strong privacy guarantees.

Main Results:

• The proposed model enables practical privacy-preserving record linkage.
• It offers controlled information leakage, avoiding the pitfalls of purely transformation-based methods.
• Empirical analysis using public voter records demonstrates the approach's practical viability.

Conclusions:

• The novel PRL model effectively balances data utility and privacy.
• It provides a more practical and secure alternative to existing methods for integrating dispersed information.
• This approach supports accurate data analysis across domains while adhering to privacy regulations.