Excessive Permissions Investigation with Data-Driven Account Security with Classification

Setiawan, Heri Satria
Pamuji, Agus
Suparman, Rudi

Jurnal Sistem Cerdas
Vol. 8 No. 2 (2025)

Many companies lack configuration systems due to the need to protect assets from unauthorized access by individuals or groups. Data mining can help by securing the configuration system to identify accounts in the database. Given the sensitivity of activities on the database system, access permissions are a major concern, especially with unauthorized users. Excessive permissions can compromise database security, making it important to group users into authorized and unauthorized classes. This study uses the decision tree method to extract and investigate factors that affect excessive permissions, and validates the dataset with 10-fold cross-validation to ensure data quality. The final result identifies two classes for user access, showing that the decision tree method performs well with significant values on the AUC curve and the Confusion Matrix