New Ways of Creating Research Data - Conversion of Unstructured Text to TEI XML using GPT on the Correspondence of Hugo Schuchard with a Web Prototype for Prompt Engineering

This paper explores the use of prompt engineering to streamline the creation of humanities research data by converting unstructured correspondence texts into the TEI XML format. The approach optimizes language models such as GPT to produce accurate structured data while preserving context. The paper discusses the iterative refinement of the conversion process, challenges, and potential solutions, and presents a user-friendly web prototype. Overall, prompt engineering shows potential for improving the efficiency of research data creation in the humanities.