Abstract:
With the widespread use of social media platforms, virality has become a fascinating subject for researchers. Content-based research emphasizes the role of content characteristics in predicting virality, such as the type of information presented, the emotional tone, and the format. On the other hand, Creator-based research looks at the characteristics of the person or group who created the content, such as their perceived credibility or popularity, as a factor in predicting virality. Both types of research can provide valuable insights into the factors that drive virality on social media. Individuals and organizations can create more effective social media campaigns and communication strategies by understanding the content and creator characteristics that promote sharing and diffusion. However, it is worth noting that many other factors contribute to virality, and the complex interplay between these factors can make it difficult to predict what will become popular online. One such factor is network-based features. We propose a topology of network features on top of content-creator features. We argue that adding the network features on top of the content-creator features will add conceptual richness and improve the predictive validity of future studies. We demonstrate this by running models, with and without the interactions, on a data set of nearly 100,000 posts from GAB, an American microblogging and social networking service known for its far-right user base. Our experiments show that our approach gives 92% F1 scores and ROC.