Research on the Fusion of Design and Business Knowledge Graph for Intelligent Decision Making System based on DSR Theory

Abstract

This paper proposes a framework of intelligent decision making system based on the fusion of Design Science Research (DSR) theory and Business Knowledge Graph (BKG), aiming at solving the complexity and uncertainty faced by the enterprise decision making in the dynamic market environment. By constructing a DSR quintuple formal model and combining it with a two-layer optimization strategy, the system realizes a full-process closed-loop design from problem definition to effect evaluation. In terms of technical implementation, an event-triggered knowledge graph dynamic expansion algorithm is proposed to quantify the strength of spatio-temporal correlation by using Haver sine formula and exponential decay function, and through the cooperative optimization of federated learning architecture (communication overhead reduced by 72ms) and hybrid inference engine (rule engine+graph neural network), the accuracy of risk prediction is improved by 28.6% (F1-score reaches 0.89), and decision response speed reduced to 1.4 seconds. The empirical study shows that the rule engine contributes 62% of the base accuracy (SHAP value of 0.41), the graph neural network recognizes the hidden correlations such as logistic delays (attention weight up to 0.72), and the model maintains its robustness under 10% noise interference (F1-score decreases by only 2.1%). The study not only provides a new theoretical paradigm for intelligent decision-making system, but also provides a real-time and interpretable decision-making tool for retail supply chain management, and will further explore the manufacturing multi-level supply chain adaptation mechanism and adaptive contingency response technology in the future.