NOMAD: A Multi-Agent LLM System for UML Class Diagram Generation from Natural Language Requirements

This article introduces NOMAD, a multi-agent LLM system designed to generate UML class diagrams from natural language requirements. The research focuses on leveraging LLMs for automated software design, specifically addressing the challenge of translating textual requirements into a visual representation. The multi-agent approach likely aims to decompose the complex task into smaller, more manageable sub-tasks, potentially improving accuracy and efficiency. The use of ArXiv suggests this is a preliminary research paper, and further evaluation and comparison with existing methods would be crucial.