"Exploring the Potential of Large Language Models in Supporting Domain Model Derivation from Requirements Elicitation Conversations"

Abstract

The thesis titled "Exploring the Potential of Large Language Models in Supporting Domain Model Derivation from Requirements Elicitation Conversations", investigates the potential of using Large Language Models (LLMs) to semi-automate the process of deriving domain models from transcripts of requirements elicitation conversations. The research focuses on evaluating the effectiveness of LLMs in generating domain models by comparing the models created by LLMs with those generated by a human modeller. The findings indicate that while LLMs can produce domain models with a high degree of agreement with human-generated models, they also introduce several unusable elements, necessitating human oversight. The thesis concludes that LLMs hold promise for enhancing the efficiency of domain modeling in requirements engineering but require careful integration to avoid errors. The study also identifies the rapid evolution of LLM technology and the context-specific nature of datasets as key limitations and recommends further research to refine the application of LLMs in this field.