Information cascade final size distributions derived from urn models

Abstract Bipolarization is a phenomenon in which either large or very small information cascade appears randomly when the retweet rate high. This phenomenon, has been observed only simulations, potential to significantly advance prediction of final sizes because forecasters need focus on two peaks size distribution rather than considering effects various details, such as network structure and user behavioral patterns. The also suggests difficulty identifying factors that lead emergence large-scale cascades. To verify existence bipolarization, this paper theoretically derives mathematical expressions using urn models, simplify diffusion behavior actual online social networks. Under assumption infinite size, exhibits power-law behavior, consistent with results existing models previous Twitter analytical outcomes. finite bipolarization observed.