Computing Node Clustering Coefficients Securely

Abstract

When performing any analysis task, some information may be leaked or scattered among individuals who may not willing to share their information (e.g., number of individual's friends and who they are). Secure multi-party computation (MPC) allows individuals to jointly perform any computation without revealing each individual's input. Here, we present two novel secure frameworks which allow node to securely compute its clustering coefficient, which we evaluate the trade off between efficiency and security of several proposed instantiations. Our results show that the cost for secure computing highly depends on network structure.