Mining Attribute-Based Access Control Policies

The flexibility feature of Attribute-Based Access Control (ABAC) makes it a powerful access control model for supporting the authorization demands of complex and dynamic systems. However, the migration from traditional access control models to the ABAC model is challenging. One promising approach to ease the burden of policy migration is policy mining. This paper proposes a bottom-up policy mining approach to automatically extract policies by mining access logs. The approach also employs machine learning techniques to learn ABAC policies. Real and synthetic data sets are employed to evaluate the approach. The experimental results demonstrate that our approach can generate ABAC policy rules efficiently.