[论文解读] Viral spread with or without emotions in online community

Andrzej Jarynowski, Jarosław Jankowski+1 more|arXiv (Cornell University)|Feb 13, 2013

Opinion Dynamics and Social Influence参考文献 21被引用 3

一句话总结

本研究将在线虚拟社区中的病毒式传播建模为流行病过程，将带有和不带激励的虚假活动与针对反ACTA条约的真实情感活动进行对比。研究发现，情感内容使二次感染概率提高四倍，表明情感显著放大了传播效应，超越单纯的激励作用。

ABSTRACT

Diffusion of information and viral content, social contagion and influence are still topics of broad evaluation. We have studied the information epidemic in a social networking platform in order compare different campaign setups. The goal of this work is to present the new knowledge obtained from studying two artificial (experimental) and one natural (where people act emotionally) viral spread that took place in a closed virtual world. We propose an approach to modeling the behavior of online community exposed on external impulses as an epidemic process. The presented results base on online multilayer system observation, and show characteristic difference between setups, moreover, some important aspects of branching processes are presented. We run experiments, where we introduced viral to system and agents were able to propagate it. There were two modes of experiment: with or without award. Dynamic of spreading both of virals were described by epidemiological model and diffusion. Results of experiments were compared with real propagation process - spontaneous organization against ACTA. During general-national protest against new antypiracy multinational agreement - ACTA, criticized for its adverse effect on e.g. freedom of expression and privacy of communication, members of chosen community could send a viral such as Stop-ACTA transparent. In this scenario, we are able to capture behavior of society, when real emotions play a role, and compare results with artificiality conditioned experiments. Moreover, we could measure effect of emotions in viral propagation. As theory explaining the role of emotions in spreading behaviour as an factor of message targeting and individuals spread emotional-oriented content in a more carefully and more influential way, the experiments show that probabilities of secondary infections are four times bigger if emotions play a role.

研究动机与目标

• 理解情感如何影响在线社区中病毒式内容的传播。
• 将带有和不带激励的虚假病毒式活动与ACTA抗议期间的真实情感活动进行对比。
• 将在线社区行为建模为流行病过程，并分析分支动力学。
• 量化情感内容对二次感染概率的影响。

提出的方法

• 在封闭虚拟世界中开展受控实验，设置两个虚假病毒式活动：一个带有激励（奖励），一个不带激励。
• 监控并记录网络中各参与方之间病毒式内容的传播动态。
• 应用流行病学模型描述传播过程，将信息扩散视为分支过程。
• 收集来自真实世界在线抗议活动的数据，用户因情感驱动而传播“停止ACTA”信息。
• 使用统计和基于网络的分析方法，对比虚假活动与情感活动的传播模式。
• 使用分支过程框架，估算每种条件下二次感染的概率。

实验结果

研究问题

• RQ1 情感的存在如何影响在线社区中病毒式内容传播的速度和广度？
• RQ2 经济激励与情感动机对信息扩散的相对影响是什么？
• RQ3 在传播动力学方面，虚假病毒式活动与真实世界情感驱动的活动相比如何？
• RQ4 流行病模型在多大程度上能准确描述社交网络中的在线信息扩散？
• RQ5 当内容分享涉及情感时，二次感染概率的定量差异有多大？

主要发现

• 与非情感或激励驱动的内容相比，情感内容使二次感染概率提高了四倍。
• 由情感驱动的真实世界ACTA抗议活动，其传播速率显著高于虚假活动。
• 激励活动虽产生可测量的传播，但其传播范围远低于情感驱动活动。
• 流行病模型准确捕捉了信息传播的动力学，尤其是在包含情感因素时。
• 分支过程模型有效描述了观察到的传播模式，各条件下传播概率存在明显差异。
• 在观察到的实验设置中，情感比物质激励成为病毒式传播的更强驱动力。