The research aims to solve the problems of information asymmetry, high credit risk and low financing efficiency in traditional supply chain financial services. This study puts forward a super-network model of supply chain finance that integrates credit information, cooperative relationship and economic activity data, and constructs a comprehensive and dynamic financial ecosystem. The model consists of three core sub-networks: credit network, cooperation network and economic network, which together form a complex and comprehensive supply chain financial framework. In this study, the graph neural network (GNN) is selected as the core deep learning algorithm to deal with data with complex relational structure, and it shows advantages in capturing dependencies between nodes, learning network structural characteristics, optimizing credit score and risk identification. Through data preprocessing, model construction, training and evaluation, the experimental results show that the research model and algorithm are superior to the traditional single machine learning model in credit score and risk identification, with higher accuracy and F1 score, and better AUC-ROC performance. This study confirms the validity and feasibility of the proposed super-network model and GNN algorithm of supply chain finance, which provides more accurate and efficient financial services for financial institutions and promotes the healthy development of supply chain finance.
Xing et al. (Sun,) studied this question.