Research on the Application of Knowledge Graph Technology in Supply Chain Network Relationship Mining and Collaborative Management

Abstract

The conventional relational mining techniques, which rely on statistical analysis or single-graph modeling approaches, exhibit several major limitations, including weak capability for multi-source heterogeneous data fusion, insufficient performance in identifying implicit relationships, and limited support for collaborative decision-making.To address these issues, this paper proposes a novel methodological framework that leverages knowledge graphs for relational mining and collaborative management in supply chain networks. Specifically, by integrating rule-based mining with deep learning methods, the framework enables a unified structured representation of multi-source data (orders, logistics, and enterprise relationships) and constructs a dynamic supply chain knowledge graph. Furthermore, graph computation and graph embedding techniques are employed to capture the topological structure of the supply chain network, while centrality analysis and link prediction are used to identify key nodes and infer latent relationships.Experimental results demonstrate that the proposed method achieves significant improvements in both relational mining and collaborative decision-making tasks. In particular, the network density increases to 0.093 while preserving the sparse characteristics of real-world supply chain networks; the average path length decreases to 3.05; the link prediction AUC reaches 0.91; and the accuracy of key node identification improves to 0.81. Regarding collaborative optimization performance, the proposed approach achieves a cost reduction of 23.5%, a time reduction of 19.6%, and a risk control improvement of 41 %.