User Item

User-item interaction modeling focuses on predicting user preferences for items, primarily to power recommender systems. Current research emphasizes improving recommendation accuracy and fairness by leveraging advanced graph neural networks (GNNs), incorporating diverse data sources like knowledge graphs and user reviews, and addressing challenges like data sparsity and cold-start problems through techniques such as contrastive learning, graph augmentation, and pre-training. These advancements aim to create more effective and robust recommender systems with broader applicability across various domains, ultimately enhancing user experience and informing business decisions.