Support ORCID and ROR reconciliation for creators and organisations

[leskneebone]

Propose that agent reconciliation targets ORCID and ror.org , at least initially.

• extracted personal creator names could optionally be reconciled against ORCID
• extracted publishers/affiliations/organisations could optionally be reconciled against ROR
• retain the literal extracted value alongside the identifier/match
• ideally expose confidence/candidate information rather than silently asserting an identity

This would implement part of the README's stated authority-control/reconciliation aspiration.

[osma]

Thank you for the suggestion. Can you suggest a concrete use case or scenario where this would be useful and important?

[leskneebone]

I guess BIBRA might produce something like:

{
  "title": "Some biodiversity report",
  "creator": ["Sam A. Name"],
  "publisher": ["Department of Environment"],
  "year": "2024",
  "type_coar": "report"
}

.. and ROR reconciliation might refine it as:

<resource-1>
    a schema:Report ;
    schema:name "Some biodiversity report" ;
    schema:creator orcid:123 ;
    schema:publisher <https://ror.org/01wsx6q69> ;
    schema:datePublished "2024"^^xsd:gYear ;
    dcterms:type <http://purl.org/coar/resource_type/c_93fc> .

Of course there are likely many organisations and people in ror.org and ORCID with those names - a confidence score based on various contextual information could be generated:

ex:123
    a ex:CandidateMatch ;
    ex:extractedValue: "Department of Environment" ;
    ex:candidateIRI "https://ror.org/01wsx6q69" ;
    ex:confidence: "70"^^xsd:decimal ;
    ex:extractionDate "2099-01-01"^^xsd:date ;
    ex:source "https://ror.org"^^xsd:anyURI .

ex:456
    a ex:CandidateMatch ;
    ex:extractedValue: "Sam A. Name" ;
    ex:candidateIRI orcid:123 ;
    ex:confidence: "55"^^xsd:decimal ;
    ex:extractionDate "2099-01-01"^^xsd:date ;
    ex:source "https://orcid.org"^^xsd:anyURI .

... and then let the application/user decide whether that becomes an asserted identity.

Where would it be useful? I guess at the workflow stage where the user gets to view candidate matches, with confidence scores.