In the era of big data, graph data has attracted considerable attention. We have witnessed the impressive performance of graph neural networks (GNNs) in dealing with graph data, as well as various real-world applications (e.g., recommender systems, molecular property prediction). The increasing number of works on GNNs indicates a global trend in both academic and industrial communities. Despite the progress made in GNNs, there are various open, unexplored, and unidentified challenges. One major concern is whether current GNNs are trustworthy. This is an inescapable problem when GNNs step into real-world applications, especially in risk-sensitive domains. To address this problem, we need to make GNNs more robust, explainable, and stable. Thus, there is a pressing demand for novel and advanced trustworthy GNNs. In this Special Issue, our goal is to bring together researchers and practitioners working in the areas of GNNs to address a wide range of theoretical and practical issues.
|