Popular Topics Spread Faster: New Dimension for Influence Propagation in Online Social Networks

Although influence propagation of Online Social Networks (OSNs) is widely studied in literature, none of the existing works can cope with the situation that the propagation rate dynamically increases for popular topics. Instead, they all assumed known rates (either constants or drawn from known distributions) for influence propagation models. However, such models cannot correctly describe influence diffusion in reality. In this paper, we propose a novel model, Dynamic Influence Propagation (DIP), that allows propagation rate to change during the diffusion, based on which we define a new problem: Threshold Activation Problem under DIP (TAP-DIP), which adds another layer of complexity over the already \#P-Hard TAP problem. Despite the hardness of TAP-DIP, we are able to design a Lipschitz Optimization scheme (LOS) that can solve TAP-DIP with $O(\log|V|)$ ratio where $V$ is the set of nodes in the OSN. Using various real OSN datasets, we experimentally demonstrate that LOS not only generates high-quality yet much smaller seed sets when being aware of the rate increase, but also scalable.