Profiling users and bots in Twitter through social media analysis

Social networks were designed to connect people online but have also been exploited to launch influence operations for manipulating society. The deployment of social bots has proven to be one of the most effective enablers to polarize and destabilize platforms. While automatic tools have been developed for their detection, the way to characterize these accounts and measure their impact is heterogeneous in the literature. In this work, we select metrics and algorithms from existing efforts to ensemble a data-driven methodology to profile groups of users and bots of Twitter from seven perspectives. We apply the framework to a dataset of Twitter retweets before the 10 November 2019 Spanish elections to characterize potential interferences. In this case study, Likely Bots (fully automated accounts) and Likely Semi-Bots (partially automated accounts) interacted with the same tendencies as Likely Humans (non-automated users), generating similar virality (information cascades) over time and without compromising the network connectivity. However, Likely Bots particularly stood out as close, visible, and reachable to other users. Likely Semi-Bots attracted particular attention, created proportionally more retweets, and were placed in strategically key positions in the core of the network. Results suggest that semi-automated accounts would be more threatening than fully automated ones. (c) 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Automated summary

This article introduces a data-driven methodology to profile Twitter users and bots from seven perspectives and applies it to retweet data before the 10 November 2019 Spanish elections to evaluate potential interferences. The results suggests that semi-automated accounts are more threatening than fully automated ones.