Published: 2025

Integrating Knowledge Graph and Large Language Model for Safety Management Regulatory Texts

CATEGORIES

RISK-BASED PROCESS SAFETY ELEMENTS

Research Summary

This conference paper combines a domain knowledge graph with an LLM to interpret and check safety‑management regulatory texts in the infrastructure industry. The authors design prompts that ground the LLM in graph‑based context, enabling more accurate compliance checking and query answering over complex regulations. Prompt patterns are tailored to tasks such as extracting obligations, mapping clauses to graph entities, and explaining compliance gaps. For PSM, the approach is directly transferable to OSHA PSM, EPA RMP, and Seveso regulations: it illustrates how prompt engineering, when coupled with structured knowledge, can support automated or semi‑automated compliance checking, regulatory cross‑referencing, and audit preparation while reducing hallucinations and

AUTHORS

Yunfei Xiang, Peng Lin, Yiming Luo, Zeyu Ning, Yuanguang Liu, Ke Liu

CITATIONS

Y. Xiang et al., “Integrating Knowledge Graph and Large Language Model for Safety Management Regulatory Texts,” in Computational and Experimental Simulations in Engineering (ICCES 2024), Mechanisms and Machine Science, vol. 175, Springer, pp. 1055–1062, first online Jan. 3, 2025.

Related Research