Dynamic graph neural network credit debt default prediction method based on collaborative supply chain

By introducing a layered sampling dynamic graph neural network model based on collaborative supply chain in credit bond default prediction, combining commodity futures and multivariate heterogeneous data, the problem of lack of forward-lookingness and difficulty in dealing with dynamic relationships in the existing technology is solved, and more accurate and robust credit risk prediction is achieved.

CN119692998APending Publication Date: 2025-03-25BEIJING INST OF TECH
0 Cites 2 Cited by

Patent Information

Application Number
CN202411507100.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-28
Publication Date
2025-03-25

AI Technical Summary

Technical Problem

The existing technology lacks forward-looking in credit bond default predictions, ignores the role of supply and demand relationships in risk transmission, and traditional graph neural networks are difficult to deal with dynamic changes in subject relationships in financial markets.

Method used

A hierarchical sampling dynamic graph neural network model (CSC-HDGCN) based on collaborative supply chain is proposed, commodity futures are introduced as risk conduction medium, combined with futures price discovery function and stock market and public opinion information, a feature cube is constructed as the initial feature of the node, and a hierarchical sampling and a uniformly distributed random sampling algorithm are adopted.

Benefits of technology

It significantly improves the accuracy of predicting corporate credit risks, provides a more comprehensive and robust risk management tool, makes up for the shortcomings of existing technologies, and provides a more forward-looking and robust technical solution for credit bond default forecasts.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a dynamic graph neural network credit debt default prediction method based on a collaborative supply chain, and belongs to the technical field of artificial intelligence and financial risk control. The invention provides a layered sampling dynamic graph neural network model based on a collaborative supply chain. The layered sampling dynamic graph neural network model is used for effectively predicting credit debt default. Commodity futures are introduced as risk conduction media, and price risks of upstream raw materials of a main industrial chain are quantified through a discovery function and convergence of futures prices. Hierarchical sampling is adopted to extract industry features, and the risk-free interest rate is utilized to accurately measure industry risks. Through a uniformly distributed random sampling algorithm, the problems of structure instability and information sparsity of the financial network are solved, and the robustness and prediction accuracy of the model are improved. According to the method, multi-dimensional data from upstream and downstream of the supply chain are effectively integrated, the prediction accuracy of the enterprise credit risk is remarkably improved, and a more comprehensive and robust risk management tool is provided.
Need to check novelty before this filing date? Find Prior Art