CDetector: Extracting Textual Features of Financial Social Media to Detect Cyber Attacks

With the proliferation of social media, cyber threats and attacks have significantly increased in complexity and quantity in financial market. Malicious hackers leverage the influence of social media to spread deceptive information with an intent to gain abnormal profits illegally or to cause losses. Measuring information content in financial social media helps identify these threats and attacks. In this paper, we propose CDetector, an ML-based approach to identifying social media features that correlate with abnormal returns of the stocks of companies vulnerable to be targets of cyber attacks (e.g., cognitive hacking). To test our approach, we collected price data and the social media messages on multiple technology companies, and extracted features that contributed to abnormal stock movements. Preliminary results show that the top social media features associated with abnormal price movements are the terms that are simple, motivate actions, incite emotion, and use exaggeration, and the selected features correlate with abnormal messages and abnormal returns of the stocks of companies.