Evaluating the Role of Social Bots in Shaping Public Opinion Online

Abstract

Social media has transformed how information is shared and has enabled both real and synthetic actors to influence the public. Automated bots are some of the most recent innovations that can be used to influence conversation, propagate false narratives, and misinformation on a massive scale. This research focuses on social bots’ impact on shaping public opinion through social media by applying a hybrid approach of bot detection using machine learning and content analysis. The competition dataset containing political debates extracted from Twitter was used to see if bots were active or human users. The study trained a logistic regression classifier with a sentiment analysis to gauge the opinion and the emotional undertone that is battered by bots. Analysis shows that bots tend to exploit polarized opinions and overemphasized sentiments, especially on trending hashtags or topics that have high participation levels. Performance comparison proved the effectiveness of the proposed detection mechanism towards bot account identification. This research calls for enhanced policies and transparency algorithms but goes further to answer the question on how social bots manipulate digital democratic activities and encourages the development of undemocratic—non-human fueled—discourse detectors.