POSTER: Dynamic Software Vulnerabilities Threat Prediction through Social Media Contextual Analysis

Abstract

Publicly available software vulnerabilities and exploit codes are often utilized by malicious actors to launch cyberattack to vulnerable targets. Therefore, organizations not only need to update their software to the latest version, they need to do effective patch management and also prioritize the patching schedule. In order to prevent future cyber threat based on publicly available resources on the Internet, this study propose a dynamic vulnerability threat assessment model to predict the exploited tendency for each vulnerability (i.e., CVE). The model considers many aspects of vulnerability which are gathered from multiple sources. Features range from profile information to contextual information of Twitter discussion about these vulnerabilities. When applied to predict the vulnerabilities exploitation in real world data, it showed better prediction accuracy using our approach and has been deployed into our threat intelligence platform as one of the analytic functions.