Christen, PeterSchnell, RainerVatsalan, DinushaRanbaduge, Thilina2024-01-3123 May 201Christen, P., Schnell, R., Vatsalan, D., Ranbaduge, T. (2017). Efficient Cryptanalysis of Bloom Filters for Privacy-Preserving Record Linkage. In: Kim, J., Shim, K., Cao, L., Lee, JG., Lin, X., Moon, YS. (eds) Advances in Knowledge Discovery and Data Mining. PAKDD 2017. Lecture Notes in Computer Science(), vol 10234. Springer, Cham. https://doi.org/10.1007/978-3-319-57454-7_499783319574530http://hdl.handle.net/1885/312471Privacy-preserving record linkage (PPRL) is the process of identifying records that represent the same entity across databases held by different organizations without revealing any sensitive information about these entities. A popular technique used in PPRL is Bloom filter encoding, which has shown to be an efficient and effective way to encode sensitive information into bit vectors while still enabling approximate matching of attribute values. However, the encoded values in Bloom filters are vulnerable to cryptanalysis attacks. Under specific conditions, these attacks are successful in that some frequent sensitive attribute values can be re-identified. In this paper we propose and evaluate on real databases a novel efficient attack on Bloom filters. Our approach is based on the construction principle of Bloom filters of hashing elements of sets into bit positions. The attack is independent of the encoding function and its parameters used, it can correctly re-identify sensitive attribute values even when various recently proposed hardening techniques have been applied, and it runs in a few seconds instead of hours.application/pdfen-AU© 2017 Springer International Publishing AGPrivacyRe-identificationFrequency analysisData linkageEntity resolutionData matching2017-04-2310.1007/978-3-319-57454-7_492022-10-02