Privacy-preserving profile matching using the social graph

We present a privacy-preserving protocol for users to test a match with potential new friends in an environment where all users cryptographically encrypt their private information. The following scenario is considered. Suppose that user Alice thinks that Bob might be a good new friend. So, Alice and the Online Social Network (representing Bob) engage in a two-party matching protocol. In this protocol no work from Bob is required, Bob can be offline. The matching protocol is designed to give Alice an indication if Bob is similar to her based on their profiles. We show that the process does so without revealing the private information of Alice and Bob to one another and to the Online Social Network.