barthanssens
commented
2 years ago Well, the dcterms:identifier could basically be anything, but do note that in RDF it is a Literal (not an URI/IRI) which may or may not look like an URI/IRI
From an academic point of view, it is also a bit superfluous, since the IRI of the catalog itself should be a unique identifier, so an identifier could be considered to be "an extra"
Anyway, I'm not so keen on using the download URL, because that might be confusing when the Catalog also describes/ links to these serializations as Dataset.
(we have a similar discussion when identifying a Dataset, and dcat:accessURL / dcat:downloadURL, which in some files generated by content platforms are exactly the same... which may lead to confusing results when quering a store with SPARQL)
So I would suggest we recommend something like java's reverse-DNS namespace (e.g. be.gov.data.catalog, I have to admit I'm not using this for data.gov.be at the moment, but I think I should) to easily create unique IDs without introducing ambiguity...
bfrichet3
Dear @barthanssens
I hope you are all right. I have a question concerning the most appropriate input in the dct:identifier attribute of dcat:Catalog classes. As you know that element is required following Belgian Federal DCAT AP Profile. Nevertheless, I don't know how to feed that field and I am looking for your best practices.
I though we could use one of the Catalog serializations' download URL (for instance the RDF/XML download URL of the catalog). Is that a good idea? What would you suggest instead of it?
Regards,
Benoît