Generating Domain-Specific Programs for Diagram Authoring with Large Language Models

Large language models (LLMs) can generate programs in general-purpose languages from prose descriptions, but are not trained on many domain-specific languages (DSLs). Diagram authoring with Penrose, a diagramming system using three DSLs, exemplifies the utility of DSL program generation with LLMs, which enables diagram creation from prose. We provide methods to conceptualize and evaluate the structures of one-shot LLM prompts to generate error-free DSL programs and implement Penrose diagram creation from prose using LLMs. We will evaluate our LLM prompt structures by testing prompt variations across different diagramming domains and plan to run a user study to assess the ease of LLM-augmented Penrose diagramming over other tools.