Privacy-Preserving Data Mining: A Game-Theoretic Approach

Privacy-preserving data mining has been an active research area in recent years due to privacy concerns in many distributed data mining settings. Protocols for privacy-preserving data mining have considered semi-honest, malicious, and covert adversarial models in cryptographic settings, whereby an adversary is assumed to follow, arbitrarily deviate from the protocol, or behaving somewhere in between these two, respectively. Semi-honest model provides weak security requiring small amount of computation, on the other hand, malicious and covert models provide strong security requiring expensive computations like homomorphic encryptions. However, game theory allows us to design protocols where parties are neither honest nor malicious but are instead viewed as rational and are assumed (only) to act in their own self-interest. In this paper, we build efficient and secure set-intersection protocol in game-theoretic setting using cryptographic primitives. Our construction avoids the use of expensive tools like homomorphic encryption and oblivious transfer. We also show that our protocol satisfies computational versions of strict Nash equilibrium and stability with respect to trembles.