MLCroissant: add fields such as creator, keyword and dates

[josvandervelde]

The current mlcroissant implementation seems to focus mostly on machine-readability. Fields such as creator are not implemented (see Metadata for the implemented fields). This makes it difficult to use the mlcroissant validation if you want to add those fields.

The Croissant specs mention that "Croissant is based on schema.org, and builds on its Dataset vocabulary."

Should the omitted schema.org fields be added to the mlcroissant implementation? Or only a select few, for instance the properties recommended by Google?

I would suggest allowing all schema.org/Dataset fields to be added when building a mlcroissant Metadata object. To keep the mlcroissant implementation small and focused, we could decide not to validate most fields - contentLocation would for example take quite some work to validate.
I'm interested in your thoughts!