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

TL;DR

This paper applies causal inference techniques to the independent cascade model in social networks, analyzing when model parameters can be identified from observational data, especially under unobserved confounding factors.

Contribution

It introduces conditions for parameter identifiability in extended IC models with unobserved confounders, a novel approach in influence propagation modeling.

Findings

Identifiability conditions for Markovian IC models

Results for semi-Markovian IC models

Analysis of IC models with global unobserved variables

Abstract

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