Application of large language models in requirements engineering

Abstract

Requirements engineering plays a significant role in software development, while the development of large language models (LLMs) provides new opportunities to support and automate various activities within this process. The aim of this paper is to analyze the application of LLMs in requirements engineering, with particular focus on the tasks in which they are applied, the methods used to evaluate their outputs and the limitations of their use. Based on a review of selected literature, the applications of LLMs in requirements elicitation, specification, analysis and validation were analyzed, along with different models, prompt formulation approaches and criteria for assessing the quality of generated outputs. The analysis indicates that LLMs can contribute to the automation of certain activities, improve requirements quality and reduce the time required for their processing, while the obtained results depend on the choice of model, task characteristics, available context and prompt formulation. Key challenges include the generation of inaccurate and biased content, context-related issues, ambiguous requirements, data protection and security concerns. Further development should focus on more reliable evaluation, improved context management and more effective collaboration between LLMs and experts.