Add provenance property to track domain data is harvested from

[AlasdairGray]

From @egonw

[For] each graph [have] a domain triple giving just the domain, e.g. for a MassBank MolecularEntity the graph would have ?g somePredicate: "massbank.eu"

From @albangaignard

You could use

• prov:wasDerivedFrom, but that needs an IRI as it is an object property
• prov:atLocation with the URL of the location

[AlasdairGray]

We could use either

• dcterms:source: "A related resource from which the described resource is derived."
• pav:providedBy: "The original provider of the encoded information (e.g. PubMed, UniProt, Science Commons)."

I think my preference would be pav:providedBy with the hostname as the value.

[AlasdairGray]

Note that when datasets come from the same site, e.g. BioSamples and Chembl both provided by https://ebi.ac.uk/, then these will be seen as coming from the same site.