Enhancing Robustness of Graph Neural Networks on Social Media with Explainable Inverse Reinforcement Learning