Causal Inference for Influence Propagation—Identifiability of the Independent Cascade Model

Independent cascade (IC) model is a widely used influence propagation for social networks. In this paper, we incorporate the concept and techniques from causal inference to study identifiability of parameters observational data in extended IC with unobserved confounding factors, which models more realistic scenarios but rarely studied modeling before. We provide conditions or unidentifiability several special structures including Markovian model, semi-Markovian global variable. Parameter important other tasks such as maximization under diffusion networks factors.