Social Bot Detection¶
Detecting automated accounts—social bots, sybils, or inauthentic agents—on social media platforms. Core problem: while most accounts are human-controlled, millions of automated agents engage in content creation and user interaction, often for manipulation, misinformation amplification, astroturfing, financial fraud, or propaganda. Detection approaches span machine learning (feature-based classifiers, neural networks) and graph-based methods (network anomaly detection), with ongoing adversarial adaptation as bot strategies evolve to evade detection.
Key papers¶
- Detecting and Tracking the Spread of Astroturf Memes in Microblog Streams — Detects bot-driven astroturfing campaigns via network topology; identifies concrete coordinated bot networks (e.g., @CStevenTucker accounts, freedomist.com bots) during 2010 elections
- Davis et al. (2016) — BotOrNot: A System to Evaluate Social Bots — Public web and API service using 1,000+ features and Random Forest for Twitter bot classification; served over 1 million requests.
- Ferrara et al. (2015) — The Rise of Social Bots — Foundational survey synthesizing detection methods, taxonomies of bot types, and impacts on information ecosystems.
- Varol et al. (2017) — Online Human-Bot Interactions: Detection, Estimation, and Characterization — Large-scale framework with 1,150 features for bot detection and population-level estimation.
- Shao et al. (2017) — The spread of low-credibility content by social bots — Empirical analysis showing bots amplify misinformation during 2016 U.S. election; 0.62 retweet ratio for low-credibility vs. 0.47 for high-credibility sources.
- Stukal et al. (2017) — Detecting Bots on Russian Political Twitter — Application of bot detection to identify state-sponsored accounts during Russian information operations.
- Yang et al. (2019) — Arming the public with artificial intelligence to counter social bots — User study on bot-detector interpretability: non-technical users struggle with raw scores; confidence calibration and explanation interfaces improve trust.
Related topics¶
- Botometer — Tool/service closely related to BotOrNot
- Misinformation Detection Methods — Bot-driven amplification is a key misinformation propagation mechanism
- Computational Propaganda — Bots used for state-sponsored information operations
- Information Spread — Bots as amplification mechanisms in information diffusion