本研究探讨了Graph Prompt作为预训练GNN模型适应下游任务的有效方法，通过任务特定的图形提示实现性能提升。提出了统一的图选择提示特征学习（GSPF）方法，增强了模型的可靠性和紧凑性。实验结果显示该方法在多个数据集上表现优越。

There are two persons in this world that helped me to gain peace in mind. Both them are alive, and not older too much than me, and ironically, both appear on the same day at the Cannes Lion to...