Discovering Latent Influence in Online Social Retweet Behaviors

The online social networks have become a powerful platform for people to communicate with each other. The study of information diffusion in social networks has attracted a lot of interest both by academia and industry. The key to modeling the diffusion is to analyze latent factors associated with the spread process. In this paper, we propose a model for discovering the latent influence and classify the latent factors into two kinds: social influence and intrinsic influence, which are the foundation of our proposed model. In the model, we present the arithmetic of the influencing parameters. Experimental results on a real dataset crawled from Sina Weibo show our model has a better accuracy than other methods. This model provides valuable insights for finding the predecessor influencer of retweet behaviors and building a more precise path of diffusion.