Time-aware denoising for neural graph collaborative filtering

This research introduces a temporal mechanism to identify and rectify noisy user-item interactions in recommendation systems. By leveraging the timing of interactions, the method purifies input data to enhance model robustness against malicious or accidental noise.