关键词: complex networks, social contagions, heterogeneity, phase transition

Abstract: Despite having significant effects on social contagions, individual heterogeneity has frequently been overlooked in earlier studies. To better understand the complexity of social contagions, a non-Markovian model incorporating heterogeneous social influence and adoption thresholds is introduced. For theoretical analysis, a generalized edge-based compartmental theory which considers the heterogeneities of social influence and adoption thresholds is developed. Focusing on the final adoption size, the critical propagation probability, and the phase transition type, social contagions for adoption thresholds that follow normal distributions with various standard deviations, follow various distributions, and correlate with degrees are investigated. When thresholds follow normal distributions, a larger standard deviation results in a larger final adoption size when the information propagation probability is relatively low. However, when the information propagation probability is relatively high, a larger standard deviation results in a smaller final adoption size. When thresholds follow various distributions, crossover phenomena in phase transition are observed when investigating the relationship of the final adoption size versus the average adoption threshold for some threshold distributions. When thresholds are correlated with degrees, similar crossover phenomena occur when investigating the relationship of the final adoption size versus the degree correlation index. Additionally, we find that increasing the heterogeneity of social influence suppresses the effects of adoption threshold heterogeneity on social contagions in three cases. Our theory predictions agree well with the simulation results.

Key words: complex networks, social contagions, heterogeneity, phase transition