This survey provides a comprehensive review of recent research on graph neural networks (GNNs) in recommender systems. With the rapid growth of online information, recommender systems play a crucial role in alleviating information overload. GNNs have gained popularity in this field due to their ability to effectively model graph-structured data, which is inherent in many recommendation scenarios. The survey categorizes GNN-based recommendation models based on the types of information used and the recommendation tasks, and discusses the challenges of applying GNNs to different types of data. It also highlights new perspectives for future research in this area. The survey collects representative papers and their open-source implementations from the GitHub repository https://github.com/wusw14/GNN-in-RS. The main contributions of this survey include a new taxonomy for GNN-based recommendation models, a comprehensive review of existing methods, and discussion of future research directions. The survey covers various aspects of GNNs in recommender systems, including graph construction, neighbor aggregation, information update, and final node representation. It also discusses the challenges and potential future directions for GNN-based recommendation systems. The survey is structured into several sections, including background and categorization, user-item collaborative filtering, sequential recommendation, and other tasks. The survey concludes with a summary of the key findings and future research directions in the field of GNN-based recommender systems.This survey provides a comprehensive review of recent research on graph neural networks (GNNs) in recommender systems. With the rapid growth of online information, recommender systems play a crucial role in alleviating information overload. GNNs have gained popularity in this field due to their ability to effectively model graph-structured data, which is inherent in many recommendation scenarios. The survey categorizes GNN-based recommendation models based on the types of information used and the recommendation tasks, and discusses the challenges of applying GNNs to different types of data. It also highlights new perspectives for future research in this area. The survey collects representative papers and their open-source implementations from the GitHub repository https://github.com/wusw14/GNN-in-RS. The main contributions of this survey include a new taxonomy for GNN-based recommendation models, a comprehensive review of existing methods, and discussion of future research directions. The survey covers various aspects of GNNs in recommender systems, including graph construction, neighbor aggregation, information update, and final node representation. It also discusses the challenges and potential future directions for GNN-based recommendation systems. The survey is structured into several sections, including background and categorization, user-item collaborative filtering, sequential recommendation, and other tasks. The survey concludes with a summary of the key findings and future research directions in the field of GNN-based recommender systems.