Abstract
Job Safety Analysis (JSA) is critical for proactively identifying workplace hazards, assessing their potential consequences, and implementing effective control measures. However, traditional JSA methods can be inefficient and prone to errors, particularly in complex industrial environments. This paper introduces AutoJSA, a knowledge-enhanced framework that leverages large language models (LLMs) to automate and optimize the JSA process. We collected 73 high-quality JSA reports from a chemical engineering company and divided the JSA workflow into three key tasks: hazard identification, consequence identification, and control measure generation. Two approaches - fine-tuning and retrieval-augmented generation (RAG) - were employed on a base LLM (GLM-4-9B-Chat) to adapt it for these domain-specific tasks. Experimental results demonstrate that both fine-tuning and RAG significantly improve task performance relative to the unmodified model, with fine-tuning generally providing larger gains. While RAG offers the advantages of dynamic knowledge retrieval and simpler implementation, it yields more modest improvements compared to fine-tuning. Moreover, a direct combination of fine-tuning and RAG did not show additional benefits under the current approach. Overall, AutoJSA effectively addresses the limitations of traditional JSA by increasing the accuracy and efficiency of safety analysis, laying the groundwork for fully automated JSA report generation. The findings underscore the potential of advanced artificial intelligence and natural language processing techniques to enhance workplace safety management in complex and rapidly evolving industrial settings.