AI & machine learning models for cloud security risk assessment
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2025
- Doi Numarası: 10.1080/07366981.2025.2575564
- Dergi Adı: EDPACS
- Derginin Tarandığı İndeksler: Scopus, ABI/INFORM, Applied Science & Technology Source, Computer & Applied Sciences, Geobase, INSPEC
- Süleyman Demirel Üniversitesi Adresli: Evet
Özet
The article introduces method of AI-supported risk modeling to identify information security risks in cloud computing environments. Risk scores were calculated based on four main frameworks. A synthetic dataset with 400 instances was created by combining 50 observations with parameters such as confidentiality, integrity, availability, probability, and impact. To label the risk levels, five supervised machine learning algorithms were trained. It was found that the best results were obtained while using COBIT and ISO 27005 frameworks. The weak performance of OCTAVE was mainly due to the complexity of its features. LIME was used to improve the model’s explainability and to locate the main decision factors that differentiate various frameworks. The results show that AI-powered models can offer accurate classification decisions with interpretable outputs that match risk methodology frameworks. The proposed framework signifies the very first stage of transitioning to scalable and explanatory decision support systems for cloud security risk assessment.