Abstract: In order to decide the appropriateness of a new user applying to join in a group in the social network, a privacy-preserving protocol was designed by the secure sum protocol and secure comparison protocol. In the privacy-preserving protocol, the secure basis protocol was devised for problems under a one-dimensional linear model, and the advanced protocol for a multi-dimensional model with a circular boundary. For the case of a single applicant and a multi-user group, the solution was converted and realized in a two-party computation model. After the proof of correctness, the complexity was discussed, and there is no leaking message during the process by the analysis of data views in each step. The privacy-preserving protocol avoids not only the blindness of auto recommendation by the net-system but also the decision delay due to the administrator's offline. In the meanwhile, the privacy of the applicant and group members can be protected without leaking any information.