Emotion-based Independent Cascade model for information propagation in online social media

Messages in online social media not only deliver factual signals but also sentiments of massive users. Moreover, emotions, especially the negative ones, can promote the propagation of information that carrying them. However, conventional information diffusion models like Independent Cascade ignore the influence of emotions. With the fact that emotions may lead to wider information propagation, we propose an emotion-based independent cascade model to describe the process of information spread in the circumstance of emotion contagion in online social media. In our model, a message is depicted as a distribution of five emotions and each emotion has a propagation probability between different users. Thus the propagation probability of a message is the weighted average of different emotions. In addition, we introduce an improved method to estimate the parameters of the proposed model. Experimental results on a realistic Weibo data set demonstrate that our model is much better in predicting the real-world information propagation than existing baselines. Our model could shed lights on the employment of emotion contagion in viral marketing of online social media.