HyMark: Application of Hybrid AI for Markdown Syntax Generation

Markdown, a markup language, faces accessibility barriers due to its moderate syntax complexity for users not so technically inclined. This paper presents a novel approach, merging rule-based and artificial intelligence (AI)-based strategies, leveraging natural language processing (NLP). By employing regular expressions, classifiers, and advanced language models, we automate the transformation of unstructured text into structured Markdown. This inclusive method enhances accessibility, making Markdown tools more user-friendly, and can rival the performance of large language models (LLMs) such as ChatGPT. Its novelty lies in the use of AI techniques combined with rule-based methods in a hybrid setting, to understand document semantics, and intelligently apply formatting, especially in optimal placements for headings and subheadings, extraction of code snippets, bullet points, and table generation. This technique minimizes the learning curve and manual effort in Markdown usage, aiding further adoption across various content creation domains, and thereby contributing to the documentation formatting practices. This manuscript will introduce and objectively compare the performance of our approach, HyMark, to that of ChatGPT.