Paper detail

EsPRESSo: Efficient Privacy-Preserving Evaluation of Sample Set Similarity

Electronic information is increasingly often shared among entities without complete mutual trust. To address related security and privacy issues, a few cryptographic techniques have emerged that support privacy-preserving information sharing and retrieval. One interesting open problem in this context involves two parties that need to assess the similarity of their datasets, but are reluctant to disclose their actual content. This paper presents an efficient and provably-secure construction supporting the privacy-preserving evaluation of sample set similarity, where similarity is measured as the Jaccard index. We present two protocols: the first securely computes the (Jaccard) similarity of two sets, and the second approximates it, using MinHash techniques, with lower complexities. We show that our novel protocols are attractive in many compelling applications, including document/multimedia similarity, biometric authentication, and genetic tests. In the process, we demonstrate that our constructions are appreciably more efficient than prior work.