NTR: add 'ORCID identifier'

[StroemPhi]

There currently is the class APOLLO_SV_00000496 (ORCID) defined as:

An identifier curated by ORCID, Inc. to denote some academic author.

Following the line of argumentation in https://github.com/ApolloDev/apollo-sv/issues/207, I propose to adopt this term into the IAO namespace with the following:

Preferred term label

ORCID iD

Synonyms

ORCID identifier, ORCID ID, ORCID id, iD, Open Researcher and Contributor ID

Textual definition

A 'centrally registered identifier' that is issued by ORCID (https://orcid.org/) and used to persistantly identify oneself as a human researcher or contributor. "You can connect your iD with your professional information — affiliations, grants, publications, peer review, and more. You can use your iD to share your information with other systems, ensuring you get recognition for all your contributions, saving you time and hassle, and reducing the risk of errors." [https://orcid.org/]

Suggested parent term

IAO:0000578 ('centrally registered identifier')

Attribution

https://orcid.org/0000-0002-1595-3213

[cthoyt]

Oh boy, more places we need to align the Bioregistry

[hoganwr]

I don't understand Charles' comment, but I support IAO taking over the class from Apollo-SV.

If these types of classes are thought to be problematic for whatever reason, then I'd like to know what modeling pattern one should use to say that an identifier is an ORCID, Twitter handle, GitHub user id, Scopus ID, UF ID, etc, etc. You could individually connect in the data every instance of an ORCID to an individual representing the ORCID "system" or "site". That would entail one extra triple per identifier...

On Fri, Oct 7, 2022 at 11:15 AM Charles Tapley Hoyt < @.***> wrote:

Oh boy, more places we need to align the Bioregistry

— Reply to this email directly, view it on GitHub https://github.com/information-artifact-ontology/IAO/issues/259#issuecomment-1271727520, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAJR55WWUJSEZPKC2U7SRA3WCA5AHANCNFSM6AAAAAAQIDQVVM . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[StroemPhi]

I'm leaving the Bioregistry comment asside to reduce the complextiy for now. (sorry @cthoyt)

I would also like to know which is the best pattern to use when saying a certain identifier is used to denote a thing.

I think the main problem one can run into is rooted in the fact that we commonly conflate the signifier with the signified in certain contexts to make things easier. If, for example, I were to use my ORCID in an annotation property as a means to express that I'm the creator of the ontology term with that annotation property (e.g. IAO:ORCID dc:creator ORCID:0000-0002-1595-3213), there is no problem and everyone would know how to interpret this triple. But if I want to say in a knowledge graph that I, an instance of NCBITaxon:'Homo Sapiens', participate in some planned process and would use my ORCID as IRI for that particular me instance, I would run into a problem, once this knowledge graph is being combined with another knowledge graph in which my instance IRI is different and there exists a triple other-me-instance-IRI IAO:'denoted by' ORCID:0000-0002-1595-3213. In this case the reasoner would complain that ORCID:0000-0002-1595-3213 cannot be a NCBITaxon:'Homo Sapiens' at the same time as also being an IAO:'identifier'.

[zhengj2007]

@StroemPhi Do you want to add 'ORCID identifier' before the release?

[StroemPhi]

@zhengj2007 yes, I can do that and would take the next free ID in the edit file, or is there a ID range table somewhere that I'm not aware of?

[zhengj2007]

I don't think IAO has any ID range table except those assigned to PNO and D-act. The next free ID should be fine.

[StroemPhi]

Regarding the question on how such identifiers should be modeled, on IAO:CRID there is the 'example of usage' annotation:

The sentence "The article has Pubmed ID 12345." contains a CRID that has two parts: one part is the CRID symbol, which is '12345'; the other part denotes the CRID registry, which is Pubmed.

I found that in OBI there are quite few identifiers subsumed under IAO:centrally registered identifier symbol instead of centrally registered identifier, but they are, missleading for the user, labeled "[...] identifier" and not "[...] identifier symbol".

If we look at OBI:digital object identifier for example, there is an 'example of usage' annotation that reads:

The doi symbol: "10.1109/5.771073" resolves to ieee website: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true&arnumber=771073

This is not true though, as the DOI symbol itself (the URI suffix), e.g. "10.1109/5.771073", needs the CRID registry part (the URI prefix), e.g. "https://doi.org/" to resolve.

So when modeling identifiers via their symbols, it feels like intorducing an unneeded layer of complexity. If I were to create a knowledge graph with the symbol pattern, should the symbol be represented with a blank node like in the following example and be instanciated in the triple store in which this data gets saved then?

doi:10.1109/5.771073 a obo:IAO_0000013 (journal article)  ;
    obo:IAO_0000235 (denoted by) [ a obo:OBI_0002110 (DOI) ;
        obo:OBI_0002815 (has representation) "10.1109/5.771073" ].

And if blank nodes are to be avoided, what should the IRI of it be then? If it were to be the full id URL, e.g. https://doi.org/10.1109/5.771073, than where is the difference to it being an IAO:identifier? And how should I avoid the clash between being the IRI for the DOI as well as for the journal article it refers to?

[zhengj2007]

@StroemPhi Since the requested term is added, the tracker need to be closed. But the issue you raised regarding CRID and CRID symbol is worth to discuss. Could you please submit a new tracker about it? Or you can submit it to OBI. Then it can be discussed in OBI call. Thanks!