The adoption of individual behavioral patterns is largely determined by influence that arrives from peers via social interactions or from other external sources. While adoption processes are commonly viewed as spreading phenomena, their outcomes depend on the actual decision mechanism of a susceptible individual. Adoptions may be induced by social influence via simple or complex contagion rules or take place seemingly in a spontaneous way without observable peer pressure. However, in reality, during the spread of a social contagion, these mechanisms may be present simultaneously by characterizing individual adoption events rather than the whole process. In turn, their mixture may induce a global spreading phenomenon that makes hinders the identity of its local driving mechanisms.
In this talk, I will first present our study on the distinguishability of the different adoption mechanisms from a microscopic view at the egocentric network level, which requires no global information about the network and the unfolding spreading process. We formulate this question as a classification problem and study it through a Bayesian likelihood approach and by using random forest classifiers. Then, I will complexify the problem with a setting in which the same node can adopt both the simple and the complex contagions in a temporal network. In that case, the distinguishability is on the contagion curves, and we aim to understand which process dominates the contagion dynamic.
Негізгі бет Distinguishing mechanisms of social contagion on networks, Elsa Andres
Пікірлер