Analyzing social and communication network structures of social bots and humans

Abstract

Recently, several journalistic accounts have suggested that Twitter is becoming a bellwether for mis- and disinformation due to the pervasiveness of bots. These bots are either automated or semi-automated. Understanding the intent and usage of these bots has piqued the scientific curiosity among researchers. To that effect, in this study, we analyze the role of bots in two distinct categories of real-world events, i.e., natural disasters and sports. We collected over 1.2 million tweets that were generated by nearly 800,000 users for Hurricane Harvey, Hurricane Irma, Hurricane Maria, and Mexico Earthquake. We corroborate our analysis by examining bots that engaged with the 2018 Winter Olympics. We collected over 1.4 million tweets generated by nearly 700,000 users based on the hashtags #Olympics2018 and #PyeongChang2018. We examined the social and communication network of bots and humans for the aforementioned events. Our results show distinctive patterns in the network structures of bots when compared with that of humans. Content analysis of the tweets further revealed that bots used hashtags more uniformly than humans, across all the events.