graph neural networks (GNNs) have gained wide popularity in recommender
systems due to their capability to capture higher-order structure information
among the nodes of users and items. However, these methods need to collect
personal interaction data between a user and the correspondin