31st European Conference on Systems, Software and Services Process Improvement, EuroSPI 2024, Munich, Almanya, 4 - 06 Eylül 2024, cilt.2179 CCIS, ss.64-79
In this research, we aim to conduct a systematic mapping study on Large Language Models (LLMs) for Software Engineering (SE). The significantly enhanced capabilities of LLMs have led to their use in many fields, including the important domain of SE. SE processes involve numerous artifacts, such as code, requirements, and documentation, which can serve as input to LLMs. To determine the potential applications of LLMs in SE, it’s crucial to understand their capabilities. Therefore, this systematic mapping study will explore the capabilities and potential of LLMs in SE tasks. Additionally, this research will address issues associated with LLMs, such as their non-deterministic nature and the problem of hallucinations. It will serve as a valuable resource for software developers, researchers, and practitioners interested in the intersection of artificial intelligence and SE, guiding their decisions on integrating these technologies.