Using LLMs for Domain-Specific Modeling Languages Development

This technical report discusses the application of Large Language Models (LLMs) in the development of domain-specific modeling languages (DSMLs) within the context of Model-Driven Engineering (MDE). It outlines the complexities involved in creating DSMLs, which require collaboration between domain specialists and technical experts. The report proposes a methodology that utilizes LLMs to support the requirement elicitation of language semantics, aiming to streamline the development process and reduce misunderstandings between stakeholders. The authors introduce a custom GPT model named MuTaGENe, designed to assist in defining model mutations, and present initial findings from its application in the context of the Wodel language. The report emphasizes the importance of rapid prototyping and continuous feedback in the early stages of DSML development, highlighting the potential of LLMs to enhance the efficiency of this iterative process. Overall, the findings suggest that LLMs can play a vital role in facilitating the creation of effective DSMLs.