Recently, machine learning-based methods, including graph neural networks (GNNs), have been widely used to establish structure–property relationships, which are crucial guidelines for materials design ...