We will likely need the Metadata section to support csv tables and .txt files that interpret (offer specific, human-readable definitions) of column headers for the dataset body. US Census datasets offer index tables that connect 'body' data in separate metadata csv files.
I'm not totally clear on what this is saying. I think it is saying we need a place to put column name definitions? This happens in the Structure section of a dataset, in the json schema.
We will likely need the Metadata section to support csv tables and .txt files that interpret (offer specific, human-readable definitions) of column headers for the dataset body. US Census datasets offer index tables that connect 'body' data in separate metadata csv files.