Crowdsourcing Cybersecurity: Cyber Attack Detection using Social Media

This article, “Crowdsourcing Cybersecurity: Cyber Attack Detection using Social Media,” presents a novel approach to identifying cyber-attacks through social media analysis, specifically Twitter. The researchers from Virginia Tech propose a framework that leverages crowdsourced data as a sensor to detect a broad range of cyber-attacks, such as distributed denial of service (DDoS) attacks, data breaches, and account hijacking, without requiring labeled samples or extensive training. Utilizing a weakly supervised method with a set of seed event triggers, the study introduces a dynamic query expansion strategy informed by convolution kernels and dependency parses to model the semantic structure of social media discourse on cyber-attacks. This methodology significantly outperforms existing methods by efficiently encoding key event characteristics and offering situational awareness into cyber events in an unsupervised manner. The findings underscore the potential of social media as a valuable resource in the cybersecurity domain, providing early detection and characterization of cyber-attacks.