Contagion in online social networks (OSN) occurs when users are exposed to information disseminated by other users. Studies of contagion are largely devoted to the spread of viral information and to local neighbor-to-neighbor contagion. However, many contagion events can be non-viral in the sense of being unpopular with low reach size, or global in the sense of being exposed to non-adjacent neighbors. This study aims to investigate the differences between local and global contagion and the different contagion patterns of viral vs. non-viral information. We analyzed three datasets and found significant differences between the temporal spreading patterns of local contagion compared to global contagion. Based on our analysis, we can successfully predict whether a user will be infected by either a local or a global contagion. We achieve an F1-score of 0.87 for non-viral information and an F1-score of 0.84 for viral information. We propose a novel method for early detection of the viral potential of an information nugget and investigate the spreading of viral and non-viral information. In addition, we analyze both viral and non-viral contagion of a topic. Differentiating between local versus global contagion, as well as between viral versus non-viral information, provides a novel perspective and better understanding of information diffusion in OSNs.
ASJC Scopus subject areas