Secure Similarity Coefficients Computation with Malicious Adversaries

Abstract

Similarity coefficients play an important role in many application aspects. Recently, a privacy-preserving similarity coefficients protocol for binary data was proposed by Wong and Kim (Computers and Mathematics with Application 2012). In this paper, we show that their protocol is not secure, even in the semi-honest model, since the client can retrieve the input of the server without deviating from the protocol. Also we propose a secure similarity coefficients computation in the presence of malicious adversaries, and prove it using the standard simulation-based security definitions for secure two-party computation. We also discuss several extensions of our protocol for settling other problems. Technical tools in our protocol include zero-knowledge proofs and distributed ElG amal encryption.