Predicting Cyber-Attacks Using Neural Language Models of Sociopolitical Events
Pechi, Daniel B.
2019
- Cyber-attacks pose an existential threat to individuals, businesses, and democracies across the world. As such, it is necessary to develop systems capable of predicting and preventing the sorts of phishing and malware attacks used to influence elections and breach private email servers like those employed prior to the 2016 presidential election. Detecting these campaigns is made intentionally ... read moredifficult by those perpetrating attacks, however there exists a growing tendency among cyber adversaries to perpetrate cyber-attacks in response to sociopolitical events. Prior work has explored using machine learning techniques to process text on social media to improve early warning systems for cyber vulnerabilities. In this paper, we propose a model which leverages news data to contextualize cyber warfare in patterns of sociopolitical events that have preceded past cyber-attacks. A variety of natural language processing and machine learning approaches are presented to model this relationship. Deep learning-based approaches proved most effective in modeling these indicators of cyber-attacks, underscoring the complex representational capacity necessary to effectively model the complex world of geopolitics.read less
- ID:
- 5q47s2193
- To Cite:
- TARC Citation Guide EndNote
