<![CDATA[Background: The main problem in cybersecurity risk management is that the risk identification process remains manual, time-consuming, and difficult to connect with technical detection results, business contexts, and information security standards. Objective: This study aims to develop a cybersecurity risk identification and management system based on the integration of machine learning and ontology using the ISO/IEC 27005:2022 approach. Methods: This study employed the Design Science Research (DSR) approach to design the Cyber Risk Intelligent System (CRIS). The system integrated machine learning algorithms, feature optimization, model interpretability, and cyber risk ontology. The classification model was developed using XGBoost with Particle Swarm Optimization (PSO)-based feature selection and was evaluated using accuracy, precision, recall, and F1-score metrics. Furthermore, an ISO/IEC 27005:2022-based ontology was used to map threats, vulnerabilities, impacts, and security control recommendations based on ISO/IEC 27002. Results: The results showed that the combination of XGBoost and PSO provided the best classification performance compared with other methods, achieving an F1-score of 0.9672, precision of 0.9661, recall of 0.9684, accuracy of 96.48%, and an ROC-AUC of 0.9891. The implementation of SHAP successfully improved model interpretability by identifying the dominant features that influenced cyber threat prediction. Ontology integration also enabled the risk identification and recommendation process to be carried out automatically, consistently, and in alignment with international standards. Conclusion: This study contributes to the development of an adaptive and structured cyber risk decision-support system that supports the automation of information security risk management in digital organizational environments.]]>
Copyrights © 2026
