The diffusion models of infections in social networks are intensively studied these last years. The existing studies concern in particular disease and rumor diffusions in social networks or financial risk in banking networks. We propose in this paper to study the diffusion problem of events within social and economic networks. In particular, we define a new problem of diffusion called the Influence Classification Problem. The objective is to find the set of nodes which are impacted by a given network. We also propose two diffusion models based on a computed threshold according to the graph and event attributes. We test our models on two real and known events : the hurricane Katrina and the fusion of Bayer and Monsanto.