Publication
GRASPER: Leveraging Knowledge Graphs for Predictive Supply Chain Analytics
Sabine Janzen; Hannah Stein; Sebastian Baer
In: Proceedings of International Workshop on AI Services and Applications (AISA’2024) at 43th International Conference on Conceptual Modeling (ER). International Conference on Conceptual Modeling (ER-2024), October 28-31, Pittsburgh, USA, 10/2024.
Abstract
Supply chain disruptions in manufacturing are increasingly prevalent due to the complexity and lack of transparency beyond direct suppliers. Early-stage disruptions often remain undetected, propagate in the network and pose significant challenges to industries reliant on component-based products such as sensors, engines, and electronics. We introduce GRASPER, an AI-based approach for detecting hidden problems in supply chains through graph-theoretic analysis of component shortages. Unlike traditional top-down market analyses, GRASPER converts Bill-of-Material (BOM) data into a semantically enriched knowledge graph in JSON-LD, incorporating historical and current market data on prices, availability, and lead times. By applying and combining graph-theoretical measures, GRASPER identifies critical components, manufacturers, and suppliers that could jeopardize production. The model's effectiveness was validated using a prototype in the sensor manufacturing industry, leveraging an open dataset from social network analysis to assess its performance in pinpointing critical nodes in the knowledge graph. This approach enhances supply chain transparency and resilience, offering significant support for manufacturers in mitigating risks and making informed decisions.