ABSTRACT

This paper investigates the dynamics of communication on social media, related to the spread of rumours, by studying the impact of micro-level agent interactions within social media discussions, on macro-level outcomes related to the diffusion of rumours. An agent-based framework is used to model social media discussions, modularly describing heterogeneous agents with differentiating characteristics, their interaction dynamics, rumour state transitions, and the evolution of networks on which these agents interact. Studying the effect of population, agent, interaction, and network characteristics, we find that some unobservable characteristics, like the initial distribution of opinions, play a significant role in rumour outcomes, particularly the homogeneity and polarisation of opinions. We report our findings on the mechanism and interactions and suggest heuristics for managers to counter the spread of unfavourable rumours.

摘要

本文通过研究社交媒体讨论中微观层面的代理人互动对与谣言传播相关的宏观层面结果的影响，研究了与谣言传播相关的社交媒体上的传播动态。基于代理的框架用于对社交媒体讨论进行建模，模块化描述具有差异化特征的异构代理、它们的交互动态、谣言状态转换以及这些代理交互的网络的演变。研究人口、代理、交互和网络特征的影响，我们发现一些不可观察的特征，如意见的初始分布，在谣言结果中发挥着重要作用，特别是意见的同质性和两极分化。