netwerk-digitaal-erfgoed / dataset-register

Components (API and crawler) for the NDE Dataset Register
https://datasetregister.netwerkdigitaalerfgoed.nl/api/
European Union Public License 1.2
4 stars 3 forks source link

Improve recursive validation #140

Open ddeboer opened 3 years ago

ddeboer commented 3 years ago

@ddeboer kun jij nog eens kijken hoe we onderstaande in SHACL kunnen opvangen (zodat de validatie goed gaat):

1. <id/dataset/nbt>  a                   schema:Dataset ;
2.        schema:identifier             <id/dataset/nbt> ;
3.        schema:mainEntityOfPage       <doc/dataset/nbt> ;
4.        schema:mainEntityOfPage       [ a                    schema:WebPage , schema:Dataset ;
                                          schema:dateModified  "2021-11-15"^^xsd:date ;
                                          schema:license       <http://creativecommons.org/publicdomain/zero/1.0/> ;
                                          schema:mainEntity    <id/dataset/nbt> ;
                                          schema:name          "Metadata beschrijving dataset NBT"@nl ;
                                          owl:sameAs           <doc/dataset/nbt>
                                      ] ;

Onze SHACL klaagt nu over de schema:mainEntityOfPage property dat er properties missen die verplicht zijn voor een schema:Dataset

Uit dataset: https://data.bibliotheken.nl/doc/dataset/nbt.ttl

Originally posted by @coret in https://github.com/netwerk-digitaal-erfgoed/register/issues/8#issuecomment-844909797

ddeboer commented 2 years ago

SHACL will match any sh:targetClass, which in our case includes schema:Dataset.

ddeboer commented 1 month ago

I’m not sure we should fix this, because the origin RDF looks wrong semantically:

/doc/dataset/nbt cannot be both a schema:WebPage and schema:Dataset as the dataset’s URI is /id/dataset/nbt, not /doc/.... They are here (incorrectly, in my eyes) linked together through an owl:sameAs.

@coret Do you think we can get this solved on the KB side?

coret commented 1 month ago

There is some movement at the KB in terms of LD publication environment and dataset descriptions. I suspect this particular problem will be solved shortly (read: months).