Advancing Requirements Engineering with Large Language Models
Claudius Ellsel, Rainer Stark · Procedia CIRP · 2025
This paper examines the application of Large Language Models (LLMs) in the requirements engineering process to improve the quality of textual product requirements. A custom-developed software tool demonstrates the utilization of LLMs in practically relevant workflows to reformulate requirements and assess their adherence to predefined quality criteria. In a comparative study, GPT-4o outperformed other models, closely aligning with human performance in quality assessments. These findings highlight the potential of LLMs to significantly enhance the process of working with textual product requirements, which often vary in quality and structure, with significant implications for the broader field of software and product development.