feat: Rate datasets when crawling

[ddeboer]

• Model ratings as schema:Ratings, with penalties subtracting from the total
  score of 100 for missing data.
• Base ratings on SHACL applied to mapped DCAT output, so improve DCAT
  SHACL where needed.
• Write output to https://data.netwerkdigitaalerfgoed.nl/registry/ratings
  named graph.

Fix #762

[coret]

I'm looking into presenting the ratings with the demonstrator. Like for https://www.goudatijdmachine.nl/omeka/api/items/13000 :

1. Why does http://www.w3.org/ns/dcat#distribution need improving?

2. As the rating is based on the DCAT version, the explanation is also given in DCAT. If the provided data description is schema.org based then these suggestions will be confusing to the dataset description provider! Do we store if the dataset description is schema.org or DCAT based? If not, can we store this property? Then we should convert the DCAT explanation to schema.org properties (in demo or crawler) if the origin was schema.org based.

[ddeboer]

1. Why does http://www.w3.org/ns/dcat#distribution need improving?

Fixed in #826.

2. As the rating is based on the DCAT version, the explanation is also given in DCAT. If the provided data description is schema.org based then these suggestions will be confusing to the dataset description provider! Do we store if the dataset description is schema.org or DCAT based? If not, can we store this property? Then we should convert the DCAT explanation to schema.org properties (in demo or crawler) if the origin was schema.org based.

I suggest to translate the DCAT keys to labels such as ‘date last modified’, ‘keywords‘, ‘language’. These are human-readable so look better in a UI and apply to both Schema.org and DCAT.