Abstract
Event-based social media sites (EBSMs) provide fertile ground for organizing and coordinating events, making them a valuable space for studying participants’ characteristics, motivations, and roles in these events. One of the most popular EBSMs is https://www.meetup.com/, from which we collected data for this study. While many social media users are human, numerous studies have demonstrated the existence of automated programs that mimic human behavior on social media—a.k.a. social bots. In this research, we aim to identify accounts that exhibit such behavior on https://www.meetup.com/ by leveraging various heuristics found in the literature and applying deep learning to assess the “botness” of these users. We also use Latent Dirichlet Allocation (LDA) topic modeling and ChatGPT to understand these bots’ topics and roles on the platform. Results show that users can be classified by their “botness” with high accuracy, that factors beyond the heuristics significantly contribute to determining “botness”, and that bots play various roles in event coordination and management.