Abrupt transitions in collaborative social networks.

Despite the wide use of networks as a versatile tool for exploring complex social systems, little is known about how to detect and forecast abrupt changes in social systems. In this report, we develop an early warning approach based on network properties to detect such changes. By analysing three collaborative social networks---one co-stardom, one patent and one scientific collaborative network, we discover that abrupt transitions inherent in these networks can serve as a good early warning signal, indicating, respectively, the dissolution of the Soviet Union, the emergence of the soft matter research field, and the merging of two scientific communities. We then develop a clique growth model that explains the universal properties of these real networks and find that they belong to a new universality class, described by the Gumbel distribution.