On-line (TweetNet) and Off-line (EpiNet): The Distinctive Structures of the Infectious

Abstract

The field of epidemic modeling has rapidly grown in recent years. However, the studies to date suffer from empirical limitations and lack the multi-layered relational investigations. The case under scrutiny is the Middle East Respiratory Syndrome (MERS) epidemic episode in South Korea 2015. Linking the confirmed MERS patients data with social media data mentioning MERS, we examine the relationship between the epidemic networks (EpiNet) and the corresponding discourse networks on Twitter (TweetNet). Using network analyses and simulations, we unpack the epidemic diffusion process and the epidemic-related social media discourse diffusion. The on-line discourse structure of the infectious is larger, dense and complex than the off-line epidemic diffusion network and it has its own unique grammar. When we differentiated tweets by their user types, we observed that they display distinct temporal and structural patterns. They showed the divergent sensitivities in the spike timing and retweet patterns compared to simulated RandomNet. High self-clustering patterns by governmental and public tweets can hinder efficient communication/information spreading. Epidemic related social media surveillance should pay customized attentions accordingly to different types of users. This study should generate discussion about the feasibility and future of liability of social media based epidemic surveillance. © 2021, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.