Cybersecurity Risk Assessment Framework for E-Governance Portals Using CVSS and Machine Learning

The increased use of e-governance portals necessitates the establishment of robust cybersecurity systems to safeguard public digital services against evolving cyberthreats. The paper presents a new approach of hybrid risk assessment approach that integrates the Common Vulnerability Scoring System (CVSS) and Machine Learning (ML) strategies. The structure has systematic procedures, including retrieving data from sources such as the NVD, preprocessing it, scoring according to CVSS, extracting features, and utilizing models like Random Forest, SVM, Neural Networks, or XGBoost. The results indicate that 42.5 percent of vulnerabilities are classified as High or Critical severity. Specifically, patching results in a 68.4% reduction in risk over six months. XGBoost was the best-performing algorithm, achieving a 95.7% accuracy score among all the experimented algorithms, with superior performance in identifying the most problematic threats, including DOS and RCE. The dashboard in the framework enables the visualization of risks and facilitates proactive choices in a real-time mode. The model is a success; however, its results are poor at low-frequency vulnerabilities and are vulnerable to limitations in CVE data. It is recommended to improve through synthetic sampling, real-time intelligence, and UI customization. On the whole, the framework provides a scalable, intelligent, and resilient model of cybersecurity, combining e-governance with technical detection methods and policy-based response options to enhance the digital infrastructure of governance.