An Attribute-Based Access Policy Mining Method Integrating Decision Trees and Clustering

With the in-depth advancement of "Internet + Government Services," the interconnection and interoperability of electronic certificates are steadily accelerating. However, as business operations become increasingly complex, traditional access control mechanisms are no longer sufficient to meet the current security requirements for accessing electronic certificates. The Attribute-Based Access Control (ABAC) model has garnered widespread attention from academia and industry due to its flexible policy expression and adaptability to complex business environments. Nonetheless, migrating policies from existing heterogeneous systems to the ABAC model and optimizing them for compatibility remain significant technical challenges. To address the issue of heterogeneous policy migration in access control, this paper proposes an attribute-based access control policy mining method that integrates decision trees with clustering. By combining the advantages of decision trees and deep learning, a deep embedded clustering method based on a multi-layer encoder driven by decision trees is introduced. This approach effectively handles the complex structures and nonlinear relationships in raw data, thereby improving clustering quality. Additionally, to address the issue of policy inaccuracy caused by redundant or incomplete rules, a policy pruning and optimization method is proposed to enhance the accuracy and usability of the extracted policies. Experimental results demonstrate that the proposed method effectively improves the accuracy of access control policy extraction.