Efficient Privacy Preserving Reputation Protocols Inspired by Secure Sum

The secure sum protocol is a well-known protocol for computing the sum of private inputs from distributed entities such that the inputs remain private. In this paper we present protocols for computing reputation in a privacy preserving manner that are inspired by the secure sum protocol. We provide a protocol that is secure under the semi-honest adversarial model as well as one that is secure under the stronger non-disruptive malicious model. Although the protocols are inspired by secure sum, they do not suffer from the issues that plague secure sum. Our protocols are resilient against colluding entities, which secure sum is not. The protocols that we develop are also efficient. We require an exchange of O(n) messages under the semi-honest model, where n is the number of feedback providers in the protocol. This is the same complexity offered by secure sum. Our protocol for the non-disruptive malicious model exchanges O(N log n) messages, where N is the number of entities in the system.