A Behavioural Analysis of Credulous Twitter Users

TL;DR

This study investigates Twitter credulous users—humans who follow many bots—by combining three public datasets into a large, labelled corpus and developing a lightweight, profile-based credulous detector using ClassA features. The authors show that simple features such as $F1$, $F14$, and $F19$ are highly informative for distinguishing credulous from non-credulous accounts and that credulous users tend to amplify bot-originated content more through retweets and replies. They validate findings with robust statistical tests (including non-parametric methods) and demonstrate practical implications for detecting misinformation spreaders and informing fact-checking efforts. The work highlights the potential of offline, feature-efficient approaches to identify susceptible audiences and understand their role in the diffusion of automated content. These insights support targeted interventions to curb spam, propaganda, and unreliable information on social platforms.

Abstract

Thanks to platforms such as Twitter and Facebook, people can know facts and events that otherwise would have been silenced. However, social media significantly contribute also to fast spreading biased and false news while targeting specific segments of the population. We have seen how false information can be spread using automated accounts, known as bots. Using Twitter as a benchmark, we investigate behavioural attitudes of so called `credulous' users, i.e., genuine accounts following many bots. Leveraging our previous work, where supervised learning is successfully applied to single out credulous users, we improve the classification task with a detailed features' analysis and provide evidence that simple and lightweight features are crucial to detect such users. Furthermore, we study the differences in the way credulous and not credulous users interact with bots and discover that credulous users tend to amplify more the content posted by bots and argue that their detection can be instrumental to get useful information on possible dissemination of spam content, propaganda, and, in general, little or no reliable information.

Figures (5)

• Figure 1: Activities of credulous users ( vs not). Each plot expresses the ratio between the tweets in the user's timeline and (a) content produced by the user, (b) retweets, and (c) replies
• Figure 2: Comparative analysis between credulous and not credulous users with respect to the retweets whose related tweets have been originated by bots.
• Figure 3: Comparative analysis between credulous and not credulous users with respect to byBot-retweets.
• Figure 4: Comparative analysis between C and NC users with respect to replies to bots' tweets.
• Figure 5: Comparative analysis between credulous and not credulous users with respect to the replies to tweets originated by bots.