DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models

DTScribe automates the creation of digital twin reporting models by ingesting digital twin documents and producing both a conceptual architecture visualization and a reporting page. It extracts key details based on a twenty‑one‑characteristic framework, generates an Ontological Modeling Language description, and refines the artifact through loops involving local large language models and ontological reasoning. The study evaluates how architectural and parameter choices—such as using medium‑sized local models and multiple refinement loops—affect artifact quality, offering practical insights for practitioners.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• llms