Automated generation of dynamical system models from text via SysML

A new strategy uses SysML diagrams and LLMs to automatically generate computational models of dynamical systems from natural language documents. The approach processes a corpus of relevant documents and an input system description through five steps, extracting component dependencies, attributes, and operations. NLP and LLMs improve intermediate outputs like key nouns, relationships, block attributes, and BDD diagrams. The method aims to accelerate engineering design and deployment.