A Derivative Topic Propagation Model Based on Multidimensional Cognition and Game Theory

Qian Li, Long Gao, Wenyi Xi, Tun Li, Rong Wang, Junwei Ge, Yunpeng Xiao

Abstract: Given that emotional content spreads more widely than rational content in social networks, as well as the complexity of user cognition and the interaction of derivative topics, this article proposes a derivative topic dissemination model that integrates multidimensional cognition and game theory. First, regarding the issue of user emotional reactions in mining topics. In this article, we quantify the affective influence among users by considering user behaviors as continuous conversations through conversation-level sentiment analysis and the proximity centrality of social networks. Second, considering that user behavior is influenced by multidimensional cognition, this article proposes a method based on S(Sensibility) R(Rationality) 2vec to simulate the dialectical relationship between sensibility and rationality in the user decision-making process. Finally, considering the cooperative and competitive relationship among derived topics, this article uses evolutionary game theory to analyze the topic life cycle and quantify its impact on user behavior by time discretization method. Accordingly, we propose a CG-back-propagation (BP) model incorporating a BP neural network to efficiently simulate the nonlinear relationship of user behavior. Experiments show that the model can not only effectively tap the influence of multidimensional cognition on users’ retweeting behavior, but also effectively perceive the propagation dynamics of derived topics.