Knowledge Graph Supply Chain Risk Management 2026
Knowledge Graph Supply Chain Risk Management
KG-SCRM applies structured graph data models to propagate and mitigate risks across complex supplier networks. Patent activity has accelerated sharply following COVID-19 disruptions and the maturation of graph neural networks and large language models.
Graph-Structured Intelligence for Supply Chain Risk
Knowledge graph-based supply chain risk management (KG-SCRM) represents supply chain entities—suppliers, products, logistics nodes, geographic regions, and financial relationships—as nodes, with typed relationships encoded as directed, weighted edges. Risk signals propagate along these structures using algorithms from probabilistic Bayesian methods to graph attention networks.
The field sits at the intersection of three technical disciplines: knowledge representation and graph databases, supply chain risk management process methodology, and AI/ML-based inference engines. Key sub-domains include static supplier network graphs, dynamic temporal knowledge graphs updated via streaming AIS data, ontology-based risk propagation models, and LLM-augmented graph reasoning.
The COVID-19 pandemic catalyzed a surge in KG-SCRM patent activity beginning in 2020. The frontier phase (2024–2026) shows three converging trends: LLM integration with knowledge graphs for natural-language critical path identification, real-time streaming knowledge graphs fed by AIS maritime data and IoT sensors, and multi-agent AI systems orchestrating KG-based risk triage.
Among retrieved records, China accounts for 14 of the filings in this dataset, making it the dominant jurisdiction for recent KG-SCRM patent activity, with the overwhelming majority of CN filings dated 2025–2026. US filings span 2018–2026 and represent more foundational platform claims. Accenture Global Solutions Limited leads with 5 filings in this dataset, followed by Siemens Aktiengesellschaft with 4.
Jurisdiction Distribution and Technology Cluster Breakdown
The 18 retrieved patent records span five main jurisdictions and four distinct technology clusters, with CN filings concentrated in 2025–2026 and US filings covering the full 2018–2026 range.
Patent Filings by Jurisdiction — KG-SCRM Dataset
In this dataset, CN filings account for 14 records, making China the most active jurisdiction, followed by US with 10 and EP and WO with 4 each.
↗ Click bars to exploreKG-SCRM Patents by Technology Cluster — Dataset Snapshot
In this dataset, the LLM and GNN-augmented inference cluster and the dynamic temporal KG cluster each represent the most active frontier areas, with foundational text-mining graph population and ontology-based propagation forming the established base.
↗ Click bars to exploreKey Application Sectors for KG-SCRM Patents
KG-SCRM patents in this dataset address six distinct application verticals, from maritime logistics and automotive supply chains to financial services and power infrastructure, each demanding domain-specific ontologies and risk propagation models.
Maritime Logistics and Global Trade
Dalian Maritime University’s 2025 CN patent targets maritime supply chains carrying over 80% of international trade, citing the 2021 Suez Canal blockage and 2023 Red Sea conflict as motivating disruption events. Inspur Zhushu Big Data’s 2026 CN filing integrates real-time AIS vessel trajectory data, port status, and supplier BOM data into a dynamic knowledge graph for continuous resilience index recalculation. Multi-level cascade risk propagation (port strike → route disruption → vessel delay → delivery failure) is addressed via dynamic graph updating.
Dynamic KG / IoT StreamingGeneral Manufacturing and Industrial Plants
Siemens’ dual 2025 EP and WO patents on automatically identifying critical objects supplied to an industrial plant assign feature vectors via graph neural network processing modules to every knowledge graph node representing materials and suppliers, raising warning tags on non-compliant nodes. IBM’s 2020 US knowledge base patent explicitly targets manufacturing workflow disruption reduction using Likert-scale geotagged disruption event classification. Sandia Corporation’s 2018 US framework applies node/edge attack-vector modeling with cost-constrained mitigation generation to government and defense supply chains.
GNN / Industrial KGAutomotive Supply Chain Risk
Hubei Mairuida Supply Chain Co., Ltd.’s 2025 CN patent introduces entropy-weighted edge scoring across tiered supplier hierarchies for automotive-specific knowledge graph risk identification. A bibliometric study covering 866 articles from 2000–2022 confirms automotive supply chain disruption risk management as a distinct and maturing research cluster. This domain requires sector-specific ontologies that capture multi-tier supplier relationships unique to automotive production networks.
Automotive KGFinancial Services and Power Infrastructure
Ping An Bank’s 2022 CN patent addresses supply chain financing risk, using knowledge graphs to enable continuous monitoring of corporate relationship graphs and business event propagation for loan approval workflows. China Electric Power Research Institute’s 2025 CN patent applies multimodal graph neural networks with graph attention mechanisms to predict equipment procurement and generation-to-consumption chain risks in the power sector. China Mobile Group’s 2025 supply chain risk assessment system integrates Monte Carlo tree search for scenario simulation of supply chain interruption credit cascades.
Finance / Energy KGLeading Patent Assignees in KG-SCRM — Dataset Snapshot
In this dataset, Accenture Global Solutions Limited holds the highest filing count with 5 records spanning US, EP, and IN jurisdictions, while Siemens Aktiengesellschaft accounts for 4 filings across EP and WO — together representing the only two assignees in retrieved records with active multi-jurisdictional prosecution strategies.
Top Assignees by Filing Count in Retrieved Records (Dataset Snapshot)
↗ Click bars to exploreAccenture Global Solutions Limited
Accenture holds 5 filings in this dataset spanning US, EP, and IN jurisdictions, covering dates from 2023 to 2025. Key patents include ontology-based risk propagation over digital twins (US and EP, 2023; US, 2025), a supply chain management platform (US, 2024; IN, 2024), and a 2025 US filing for assigning multi-agent AI systems to prioritized supply chain risk use cases. The ontology-based digital twin architecture aggregates direct asset-node risk and indirect process-chain propagated risk, with the 2025 agent orchestration patent extending this to autonomous AI triage — all maintaining active legal status.
United States / Europe / IndiaSiemens Aktiengesellschaft
Siemens holds 4 filings in this dataset across EP and WO jurisdictions, all dated October 2025. Two WO/EP counterpart patents introduce a graph-to-text encoder that converts supply chain knowledge graph structure into natural language prompts for an LLM, which returns critical path predictions with explanatory statements. Two further EP and WO patents use a graph neural network processing module to assign feature vectors to knowledge graph nodes representing materials and suppliers, raising warning tags on non-compliant nodes — representing active multi-jurisdictional prosecution signaling high perceived global commercial relevance.
Germany — EP / WOEmerging Technical Trends in KG-SCRM (2025–2026)
Five frontier directions are identifiable from filings dated 2025–2026 in this dataset, spanning LLM-KG coupling, agentic AI orchestration, real-time AIS streaming, event-theoretic graph models, and counterfactual reasoning modules.
LLM-Knowledge Graph Coupling via Graph-to-Text Encoders
Siemens’ dual WO and EP patents from October 2025 introduce a graph-to-text encoder that converts KG structure into natural language prompts, which an LLM then processes to output critical paths and explanatory statements. This represents a fundamental architectural shift: KGs previously served as query-time data stores and now serve as structured context injected into generative AI inference. R&D teams must prioritize preserving graph semantics through the natural language translation step, as information fidelity loss is the primary technical risk in this pipeline.
Agentic AI Orchestration over Supply Chain Knowledge Graphs
Accenture’s June 2025 US filing assigns multi-agent AI systems to prioritized risk use cases based on agent performance scores, representing convergence of KG-SCRM with autonomous agent architectures. This extends the digital twin ontology-based risk propagation model into a fully orchestrated, self-directing system for triage of supply chain risk events. The patent covers the assignment logic that routes specific risk use cases to the best-performing agent within the knowledge graph-backed system.
Static vs. Dynamic Knowledge Graph Architectures for Supply Chain Risk
Click any row to explore further.
| Dimension | Static / Batch KG Approach | Dynamic / Streaming KG Approach |
|---|---|---|
| Update Frequency | Periodic batch updates from ERP, procurement records, corporate filings | Continuous second-level updates via AIS vessel streams, IoT sensors, real-time news feeds |
| Representative Patents | Refinitiv Risk identification engine (US, 2018, 2022); IBM Cognitively-Derived Knowledge Base (US, 2020) | Inspur DeltaGraph incremental KG (CN, 2025); Inspur Zhushu AIS-integrated risk perception (CN, 2026); Shanghai Jujun SEIR risk propagation (CN, 2026) |
| Core Graph Algorithm | Text mining NLP, weighted criticality/centrality/distance metrics, probabilistic decision paths | Graph-BERT entity alignment, DynamicLouvain temporal community detection, graph attention network risk propagation |
| Primary Risk Model | Likert-scale disruption classification, Bayesian decision-making, node/edge attack-vector scoring | SEIR epidemic model adapted for risk propagation, multi-dimensional resilience indices, cascading failure stress testing |
| Scale Addressed | Enterprise supplier networks; direct and transitive risk across known supplier graphs | Million-node graphs with second-level update latency; global port and shipping route coverage |
| Jurisdictional Concentration | US filings predominant (2018–2022); larger enterprise platform assignees | CN filings predominant (2025–2026); mix of universities, state research institutes, and private SMEs |
| LLM Integration | Not present in foundational filings; knowledge graph queried via structured graph traversal | Siemens 2025 graph-to-text-to-LLM pipeline injects KG structure as natural language context for generative AI inference |
Frequently Asked Questions: Knowledge Graph Supply Chain Risk Management
KG-SCRM represents supply chain entities—suppliers, products, logistics nodes, geographic regions, and financial relationships—as nodes in a graph, with typed relationships (supply, transport, contractual, geopolitical dependency) encoded as directed, weighted edges. Risk signals are then propagated along these structures using algorithms ranging from probabilistic Bayesian methods to graph attention networks, enabling identification and mitigation of risks across complex supplier networks.
In this dataset of 18 retrieved utility patents, Accenture Global Solutions Limited holds the highest count with 5 filings (US, EP, IN), followed by Siemens Aktiengesellschaft with 4 filings (EP, WO). QOMPLX LLC, Strong Force VCN Portfolio 2019 LLC, Procore Technologies, and Gas Technology Institute each hold 3 filings in this dataset.
Siemens’ October 2025 WO and EP patents introduce a graph-to-text encoder that converts the supply chain knowledge graph structure into natural language descriptions. An LLM is then prompted with this description to output critical path predictions and explanatory statements. This is described as a new human-AI interaction model where the knowledge graph serves as structured context injected into generative AI inference, rather than a query-time data store.
Dynamic/temporal KGs continuously update graph structures using live data feeds such as AIS vessel tracking, IoT sensors, social media, and news streams, enabling near-real-time risk state estimation. Inspur’s 2025 CN patent uses DeltaGraph incremental graph computing for million-node second-level updates. In contrast, static approaches rely on periodic batch updates from ERP systems, procurement records, and corporate filings, using text mining and weighted centrality metrics.
In this dataset, the most active sectors are general manufacturing and industrial plants (targeted by Siemens and IBM), maritime logistics (Dalian Maritime University and Inspur Zhushu, 2025–2026 CN), automotive supply chains (Hubei Mairuida, CN, 2025), financial services supply chain finance (Ping An Bank, CN, 2022), power and energy infrastructure (China Electric Power Research Institute, CN, 2025), and construction (Procore Technologies, US, 2023–2026).
Hebei Xiong’an Xinruo Technology’s 2025 CN patent introduces a counterfactual inference module that simulates specific disruptive events within a knowledge graph, modeling low-probability/high-impact scenarios and generating backup response plans. This is described as a nascent but significant capability for tail-risk management, and represents a shift from predominantly reactive KG-SCRM systems (risk detection and propagation scoring) toward proactive scenario planning—relevant for regulatory compliance under EU supply chain due diligence frameworks and US executive orders on critical supply chain resilience.
Data and insights on this page are based on a limited patent and literature dataset and are for reference only. Figures may not represent the complete technology landscape.