Assessing Cybersecurity Threats: The Application of NLP in Advanced Threat Intelligence Systems

Cybersecurity involves safeguarding critical infrastructure and sensitive data from attackers. Various sectors, including government organizations, banks, hospitals, and other industries, are progressively allocating resources to enhance their cybersecurity infrastructure to protect their operations and the vast number of consumers who rely on them for their data. In a more interconnected world, businesses are seeing significant levels of cyber threat activity, which raises questions about their ability to defend against widespread attacks. Threat intelligence systems utilize Natural Language Processing (NLP) to analyze words and technical data in multiple languages to identify trends and patterns. This study seeks to create a system that focuses on detecting software vulnerabilities by treating source code as texts and applying powerful deep-learning NLP models. In addition, different deep learning models have been assessed as well as compared based on their accuracy. This study indicates that the CodeBERT model excels at identifying and classifying software vulnerabilities in code with the top performance achieving 95% accuracy. Furthermore, robust dashboard utilizing FastAPI and ReactJS have been created considering the vulnerability class to which a specific piece of source code belongs.