Record linkage is the process of identifying which records in different databases refer to the same real world entities. When personal details of individuals, such as names and addresses, are used to link databases across different organisations, then privacy becomes a major concern. Often it is not permissible to exchange identifying data among organisations. Linking databases in situations where no private or confidential information can be revealed is known as ‘privacy-preserving record linkage’ (PPRL). We propose a novel protocol for scalable and approximate PPRL based on Bloom filters in a scenario where no third party is available to conduct a linkage. While two-party protocols are more secure because there is no possibility of collusion between one of the database owners and the third party, these protocols generally require more complex and expensive techniques to ensure that a database owner cannot infer any sensitive information about the other party’s data during the linkage process. Our two-party protocol uses an efficient privacy technique called Bloom filters, and conducts an iterative classification of record pairs into matches and non-matches, as selected bits of the Bloom filters are revealed. Experiments conducted on real-world databases that contain nearly two million records, show that our protocol is scalable to large databases while providing sufficient privacy characteristics and achieving high linkage quality.
|Cite as: Vatsalan, D. and Christen, P. (2012). An Iterative Two-Party Protocol for Scalable Privacy-Preserving Record Linkage. In Proc. Data Mining and Analytics 2012 (AusDM 2012) Sydney, Australia. CRPIT, 134. Zhao, Y., Li, J. , Kennedy, P.J. and Christen, P. Eds., ACS. 127 - 138 |
(local if available)