Abstract: Despite advancements using graph neural networks (GNNs) to capture complex user-item interactions, challenges persist due to data sparsity and noise. To address these, self-supervised ...