Large Language Model for Requirements Engineering: A Systematic Literature Review

Large language models (LLMs) have been applied to various domains, including software engineering (SE), with comparatively better success. Recently, researchers have started exploring the capabilities of LLMs in requirements engineering (RE) by proposing various approaches for automating RE activities. Moreover, a comprehensive understanding of LLM application in various RE-related activities is still emerging and growing. For this purpose, we conducted a systematic literature review (SLR) by exploring LLMs applications in RE (LLM4RE), aiming to understand how LLMs can help improve various RE activities, such as requirements elicitation, analysis, modelling, validation, specification, prioritization, and tracing. Using inclusion and exclusion criteria to answer various research questions, we identify and critically analyze 35 research papers from 2023 to October 2024. In particular, we are interested in answering LLMs application in various RE activities, identifying various LLM strategies to train LLMs for particular RE activities, determining what different evaluation matrices are used to evaluate LLMs performance, and determining which RE activities have been effectively answered to date. Through the proposed research questions, we identified that several RE activities have been comparatively improved and possibly automated. Also, various evaluation metrics have been identified that effectively evaluate the outputs generated by the LLMs. The study can prove to be effective for software vendors and researchers in automating various RE-related activities for improved software quality and user satisfaction.