Diffusion Structure Inference under Heterogeneous Network Cascade

a cascade over a network refers to the diffusion process where behavior changes occurring in one part of an interconnected population lead to a series of sequential changes throughout the entire population. In recent years, there has been a surge in interest and efforts to understand and model cascade mechanisms, as they motivate many significant research topics across different disciplines. The propagation structure of cascades is governed by the underlying diffusion networks, which are often hidden. Inferring diffusion networks enables interventions in the cascading process to maximize information propagation and provides insights into the Granger causality of interaction mechanisms among individuals. In this project, we propose a novel double mixture graphical model for inferring latent diffusion networks in the presence of strong cascade heterogeneity. The new model represents cascade pathways as a distributional mixture over diffusion networks, where these networks capture different cascading patterns at the population level. We develop a data-driven optimization method to infer diffusion networks using only visible temporal cascade records, without requiring the modeling of complex and heterogeneous individual states. Both statistical and computational guarantees for the proposed method are established. We apply the double mixture cascade model to analyze the research topic cascades in social science across U.S. universities and uncover the latent research topic diffusion networks among the top U.S. social science programs. This is joint work with Siyu Huang and Abdul Adeel from PSU.