<p>Graph Neural Networks (GNNs) have shown strong performance in graph analysis tasks such as node classification. However, they are often treated as black boxes, which limits their adoption in high-stakes domains. Existing explanation methods suffer from trade-offs between representational capacity and interpretability, and their outputs are often insufficiently intuitive for users to efficiently gain insights. While visual analytics tools have been successfully applied in explainable deep learning, little research has addressed the unique challenges of GNN explainability. In this paper, we present a novel visual analytics framework that integrates an explanation method with interactive visualization to support the exploration of local explanations for GNNs. The proposed explanation method produces both feature- and rule-based explanations for node predictions, while the visualization system enables multi-perspective analysis to understand, diagnose, and improve GNN predictions. Quantitative experiments on real-world datasets demonstrate that our explanation method generates more expressive explanations compared to existing methods, and two case studies on real-world datasets further validate the effectiveness of the system in facilitating model interpretability and refinement.</p> Graphical abstract <p></p>