`describegpt`: Use Semantic Markdown (WIP)

As we focus on "Opening Data as a Verb": > Opening Data as a Verb is **Humanizing and Contextualizing Data** > by **adding High Resolution, High Quality FAIR metadata** > to make it **Useful, Usable and Used** > by **Human, Agents and Machines** alike. We need to make our Markdown more reliably interpretable by Agents. For Markdown has become the de facto agent format - skills, prompts, tools are primarily saved in Markdown. This is to be celebrated! For before, Humans were forced to think like Machines to communicate with Machines. In the very beginning, we were physically rewiring machines to program them. Then came Assembly language, and then over time, more readable languages like C, Pascal, Java - with which we built Frameworks and Libraries - layering abstraction over abstraction. Still, even today, modern formal specification languages still require us to think like Machines - OWL, Croissant, JSON Schema, etc. But now, the pendulum has swung to the other extreme - Plain English in Markdown format! This leads to unreliable Agent Interpretation. Semantic Markdown seeks to be the middle ground between freeform English and Formal Specification Languages. We do this by adding special annotations to Markdown to indicate Concepts and Relationships in a way that is intuitive and readable to non-technical Humans, while giving Agents more concrete guidance. Much the same way that the `@<FILENAME.md>` has emerged as a way to include other Markdown files, we add these small Markdown conventions. - TYPE/CLASS Type/class are denoted by a Heading symbol (pound sign) with the type/class name starting with a `*` prefix. For example, this denotes a `Data Dictionary` class<br>`# *Data_Dictionary`.<br>In the same Markdown file is a `# *GLOSSARY` followed by a list (ordered or unordered) of Types/Classes with their Definitions., e.g.:<br> > # *GLOSSARY > * *Data_Dictionary - describes the schema of a dataset. It's a markdown table. The header row starts with the field name, its data type, a human-readable label, and Description. Additional columns describe different characteristics of the field. There is an entry in the table for each field in the dataset. > * *Description - a description of the dataset. > * *Tags - helps classify the dataset. > * *Next Type - a type description > * *A_Class - a class description - PROPERTY cc @wardi Also see https://hackmd.io/@sparna/semantic-markdown-draft