Aiming at producing more realistic and informed agent-based simulations of social systems, one often need to build realistic synthetic populations. Apart of this synthetic population generation, the question of generating realistic social networks is an important phase. We examined the articles published in the Journal of Artificial Societies and Social Simulation (JASSS) in between 1998 and 2015 in order to identify the models of social networks that were actually used by the community. After presenting the main models (regular networks, random graphs, small-world networks, scale-free networks, spatial networks), we discuss the evolution of the use of each one of these models. We then present different existing alternatives to those kind of models and discuss the combined use of both simple and more elaborated or data-driven models to different aims along the process of developing agent-based social simulation with realistic synthetic populations