Graph Neural Networks for Collaborative Filtering: A Survey on Ranking Prediction

Recommender systems have undergone more than three decades of continuous development, from early collaborative filtering techniques and matrix factorization to the integration of deep learning methods. In recent years, ranking prediction in collaborative filtering has attracted particular attention, as it directly impacts personalization quality in the domains such as e-commerce, social media, and multimedia services. Among various approaches, Graph Neural Networks have emerged as a key paradigm due to their ability to model high-order user–item relationships and capture rich contextual information. This survey provides a systematic and up-to-date reviewof graph-based collaborative filtering models for ranking prediction, covering the period 2020–2025 and analyzing eighty representative works. We classify existing methods along multiple dimensions, including graph learning techniques, model architectures, auxiliary learning strategies, recommendation scenarios, and application domains. Furthermore, we summarize evaluation protocols and benchmark datasets to provide practical guidance for future experimentation. Finally, we discuss major challenges—such as data sparsity, cold start, oversmoothing, fairness, and scalability—and outline promising research directions, including self-supervised learning, multimodal modeling, and the integration of large language models into recommender systems. This work aims to serve both researchers and practitioners as a comprehensive reference and a roadmap for advancing the field of GNN-based ranking prediction in collaborative filtering.