Circle-Based Tipping Cascades in Social Networks

Traditional models of cascades in social networks are based on the idea that individuals become “activated” once a certain number of their friends are active. However, some recent work has suggested that in some cases individuals become active based on the receipt of social signals from circles of their friends. In this paper, we introduce a deterministic model of this circle-based activation process and study the problem of finding a minimal subset of initially-activated individuals to produce a cascade that activates the entire population. We design decomposition-based heuristics for this problem and experimentally evaluate them on various simulated and real-world datasets. Our empirical results not only provide insight into the development of heuristics for this problem, but also shed light on the effect the number of circles has upon the cascades.