Why so Emotional? An Analysis of Emotional Bot-generated Content on Twitter

Abstract

In this paper, we present a study on the emotions conveyed in b ot-generated Twitter messages as compared to emotions conveyed in human-generated messages. Social b ts are software programs that automatically produce messages and interact with human users on social med ia platforms. In recent years, bots have become quite complex and may mimic the behavior of human users. Prio r studies have shown that emotional messages may significantly influence their readers. Therefore, it is i mportant to study the effects that emotional botgenerated content has on the reactions of human users and on i formation diffusion over online social networks (OSNs). For the purposes of this paper, we analyzed 1.3 milli on Twitter accounts that generated 4.4 million tweets related to 24 systematically chosen real-world even ts. Our findings show that: 1) bots emotionally polarize during controversial events and even inject polar izing emotions into the Twitter discourse on harmless events such as Thanksgiving, 2) humans generally tend to con f rm to the base emotion of the respective event, while bots contribute to the higher intensity of shifted emo tions (i.e. emotions that do not conform to the base emotion of the respective event), 3) bots tend to shift emoti ons to receive more attention (in terms of likes and retweets).