error in nanopub typing for reference publication

[micheldumontier]

Looks like there is a problem in the nanopub in reference to the publication, namely the use of the predicate dct:type has a literal value of "dctypes"

    biolink:publications <https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=c0c3eeb6-8a75-0b20-2008-396e63cddcdb> .
  <https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=c0c3eeb6-8a75-0b20-2008-396e63cddcdb> dct:type "dctypes" ;
    a biolink:Publication ;

from https://np.petapico.org/RAqsy_jgx13zmcoR2I10ir6LEjupK2UmsWyA3DXwV7ZTE

[vemonet]

It is not a bug, it is a feature!

The BioLink model required to define a type for publication, and was using dct:type with cardinality 1..1, pointing to a restricted string, which should be one of:

 values_from: 
    - dctypes
    - fabio
    - MESH_PUB
    - COAR_RESOURCE
    - WIKIDATA 

More about it in the Publication doc page: https://biolink.github.io/biolink-model/docs/Publication.html

Description: Ontology term for publication type may be drawn from Dublin Core types (https://www.dublincore.org/specifications/dublin-core/dcmi-type-vocabulary/), FRBR-aligned Bibliographic Ontology (https://sparontologies.github.io/fabio/current/fabio.html), the MESH publication types (https://www.nlm.nih.gov/mesh/pubtypes.html), the Confederation of Open Access Repositories (COAR) Controlled Vocabulary for Resource Type Genres (http://vocabularies.coar-repositories.org/documentation/resource_types/), Wikidata (https://www.wikidata.org/wiki/Wikidata:Publication_types), or equivalent publication type ontology. When a given publication type ontology term is used within a given knowledge graph, then the CURIE identified term must be documented in the graph as a concept node of biolink:category biolink:OntologyClass.

It was actually creating issues with the SHACL validation, but the pull request to fix it has been merged recently, so that should not be a problem once I update the Collaboratory to the latest version of the BioLink SHACL shape: biolink/biolink-model#1218

To pass SHACL validation Collaboratory automatically added dctypes, because the BioLink SHACL shape contained:

[ sh:description "Ontology term for publication type may be drawn from Dublin Core types (https://www.dublincore.org/specifications/dublin-core/dcmi-type-vocabulary/), FRBR-aligned Bibliographic Ontology (https://sparontologies.github.io/fabio/current/fabio.html), the MESH publication types (https://www.nlm.nih.gov/mesh/pubtypes.html), the Confederation of Open Access Repositories (COAR) Controlled Vocabulary for Resource Type Genres (http://vocabularies.coar-repositories.org/documentation/resource_types/), Wikidata (https://www.wikidata.org/wiki/Wikidata:Publication_types), or equivalent publication type ontology. When a given publication type ontology term is used within a given knowledge graph, then the CURIE identified term must be documented in the graph as a concept node of biolink:category biolink:OntologyClass." ;
            sh:maxCount 1 ;
            sh:minCount 1 ;
            sh:order 14 ;
            sh:path dct:type ],

But checking at the current state of the SHACL shape it seems like the cardinality for dct:type has changed to publication type 0..1, so we can probably remove it now

[micheldumontier]

Right, the text is very clear that it should be a CURIE for a type that is defined in one of those terminologies, not that it should be one of those terminologies. It's good that you caught the problem and that this will be fixed for the SHACL validation.