HTAN: An Automatic Classification Method for Access Control Entities Based on GNN

In the context of big data, access control technologies have demonstrated significant importance in preventing unauthorised access and safeguarding users' sensitive information and critical resources from misuse. However, traditional access control methods, which rely on static rules and permission configurations, struggle to address the increasingly complex dynamic environments and diverse access requirements. Nowadays, graph neural network technology is increasingly demonstrating excellent performance in the field of access control and has been widely adopted across various fields. This paper proposes a heterogeneous graph neural network based on a hierarchical attention mechanism and Transformer architecture, including feature selection, node-level attention mechanism, semantic-level attention mechanism, position encoding, multi-head self-attention mechanism, and feedforward neural network. This ensures that entity classification simultaneously considers the global structure and local details of graph data, thereby improving performance and efficiency in entity classification on heterogeneous graph data. The experimental results prove that our model has excellent performance in heterogeneous graph data entity classification and also demonstrates good interpretability in graph analysis.