A Graph-Retrieval-Augmented Generation Framework Enhances Decision-Making in the Circular Economy

2025-06-01

AI Summary

Large language models (LLMs) exhibit hallucinations—particularly concerning industrial coding standards and greenhouse gas emission factors—undermining regulatory compliance and investment decisions in sustainable manufacturing. Method: This study constructs a domain-specific knowledge graph (KG) aligned with circular economy principles, pioneering the tight integration of structured KGs with retrieval-augmented generation (RAG). The framework enables SPARQL-driven, auditable multi-hop reasoning and verification subgraph retrieval, incorporating KG construction, SPARQL query translation, GWP100 emission data integration, and multi-hop question-answering inference. Contribution/Results: Experiments demonstrate perfect factual fidelity (ROUGE-L F1 = 1.0 vs. baseline <0.08), 50% reduction in response latency, and 16% lower token consumption. The approach significantly enhances factual accuracy, auditability, traceability, and regulatory readiness—establishing a trustworthy generative paradigm for green manufacturing decision-making.

Technology Category

• Machine Learning: Efficient ML / Green AI
• Knowledge Representation and Reasoning: Knowledge Representation Languages
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Large language models (LLMs) hold promise for sustainable manufacturing, but often hallucinate industrial codes and emission factors, undermining regulatory and investment decisions. We introduce CircuGraphRAG, a retrieval-augmented generation (RAG) framework that grounds LLMs outputs in a domain-specific knowledge graph for the circular economy. This graph connects 117,380 industrial and waste entities with classification codes and GWP100 emission data, enabling structured multi-hop reasoning. Natural language queries are translated into SPARQL and verified subgraphs are retrieved to ensure accuracy and traceability. Compared with Standalone LLMs and Naive RAG, CircuGraphRAG achieves superior performance in single-hop and multi-hop question answering, with ROUGE-L F1 scores up to 1.0, while baseline scores below 0.08. It also improves efficiency, halving the response time and reducing token usage by 16% in representative tasks. CircuGraphRAG provides fact-checked, regulatory-ready support for circular economy planning, advancing reliable, low-carbon resource decision making.

Problem

• Reduces LLM hallucinations in industrial codes and emissions
• Enhances decision-making in circular economy via knowledge graphs
• Improves accuracy and efficiency in sustainable manufacturing queries

Innovation

• Graph-RAG framework for circular economy
• Domain-specific knowledge graph integration
• SPARQL-based query translation and verification