add ROR as an Author Identifier Type

[pdurbin]

What this PR does / why we need it:

This PR adds ROR as an Author Identifier Type.

Which issue(s) this PR closes:

• Closes Feature Request: Add ROR to the citation metadata block author identifier field #11075

Special notes for your reviewer:

I have a question. How can a ROR uniquely identify an author? If I use the ROR for Harvard ( https://ror.org/03vek6s52 ), how can someone know which of the thousands of Harvard affiliates I am? 🤔

Also, is it ok that ROR has been added to the bottom of the list? That's what we've always done before. Here's how it looks:

Nevermind. This was fixed in ddb6bc9. ROR is now after ORCID:

Suggestions on how to test this:

Does this PR introduce a user interface change? If mockups are available, please link/include them here:

Yes, see screenshots above.

Is there a release notes update needed for this change?:

Yes, coming.

Additional documentation:

[qqmyers]

The intent is to allow Organizations to be authors. As you say, it would not make sense to use a ROR for a person. Eventually an external vocab script would start by letting you select Person or Org, but until then, perhaps @jggautier might want changes to the title or help tip to make things clearer?

[pdurbin]

@qqmyers yes, good idea. I do think the tooltip should be edited to account for organizations as authors.

@jggautier do you want me to take a whack at that? Or do you want to suggest something?

[jggautier]

Hey @pdurbin

I have a question. How can a ROR uniquely identify an author? If I use the ROR for Harvard ( https://ror.org/03vek6s52 ), how can someone know which of the thousands of Harvard affiliates I am? 🤔

Hmmm, a ROR can uniquely identifier an author when that author is an organization, such as Harvard University.

When you ask "how can someone know which of the thousands of Harvard affiliates I am?", this makes me think that you think that the Identifier Type and Identifier fields are supposed to be used for the identifier of the author's organizational affiliation, in your example Harvard University. Is that right?

It's telling that you think that, because others might, too. During testing after implementation I'll make sure we watch out for this.

Taking your example, we're expecting that if a depositor selects ROR as the Identifier Type, that means that the author is Harvard University, not that the author's affiliation is Harvard University.

Also, is it ok that ROR has been added to the bottom of the list? That's what we've always done before.

Could you write a bit more about we've always done that? I imagine there are some cases where we think it'll be better to list options in the list alphabetically or by how often we see people selecting them. Or in this case, since we want to encourage the use of ROR over other types of identifiers used for organizations, in the mockup at #11075 I added it below ORCID.

[pdurbin]

@jggautier I think we wrote at the same time. 😄 Do you want me to try editing the tooltip? Right now it only applies to people, not organizations:

[jggautier]

Ah, I'm not sure how the Identifier Type field's tooltip applies to people and not organizations. Could you write why you think that? Or are you referring to another tooltip?

[pdurbin]

I'm saying we could change...

"The type of identifier that uniquely identifies the author (e.g. ORCID, ISNI)"

...to something like this:

"The type of identifier that uniquely identifies the author as a person (e.g. ORCID, ISNI) or organization (e.g. ROR)"

[jggautier]

Ah I see. So you think that we should note in the tooltip that the author can be a person or an organization because other folks might not realize that they can add an identifier for organizational author? I think we should keep an eye on this as we get more feedback and that we shouldn't change the tooltip in this PR.

And kind of related, in our metadata text guidelines we say not to "include examples in the tooltip text for dropdown fields, since the dropdown fields already include all options". That's the 13th thing on the list. I'm surprised now that this wasn't done when we applied the guidelines to most of the fields in the citation metadata block a few years back as part of #8127.

[coveralls]

coverage: 22.7%. remained the same
when pulling 0d5665c on 11075-ror
into 75ba6d6 on develop.

[pdurbin]

I'd say I go in thinking "Author... what is an author?... it's a person... like the author of a book... or an author of an article. I see that my name is auto-populated as the author."

"Affiliation makes sense. What's next?" I think.

I get down to "Identifier Type" and see "The type of identifier that uniquely identifies the author (e.g. ORCID, ISNI)".

I think, "ORCID, yes, this will uniquely identify an author." (Again, I'm thinking of an author as a person.)

Then I see "ROR" in the list and I think "a ROR cannot uniquely identify a person. Why is ROR in this list of identifier types? Is this a bug?"

The "Author", "Name", and "Affiliation" tooltips all help reduce confusion by stating that an author can be an organization but the "Identifier Type" tooltip does not. So, I'm simply suggesting we fix this, not necessarily in this pull request, but at some point.

[jggautier]

Thanks for sharing more of your thinking @pdurbin.

I've wondered too if managers of other Dataverse installations have also felt or learned that their users feel that "Author" always means "person" and never "organization, and if that's any part of the reason why some installations use the word "Creator" instead. We've seen something similar with the Related Publication field label, so we adjusted the tooltip there a while back.

I definitely plan to help with learning about and clearing up the confusion.

[pdurbin]

Also, is it ok that ROR has been added to the bottom of the list? That's what we've always done before.

Could you write a bit more about we've always done that? I imagine there are some cases where we think it'll be better to list options in the list alphabetically or by how often we see people selecting them. Or in this case, since we want to encourage the use of ROR over other types of identifiers used for organizations, in the mockup at #11075 I added it below ORCID.

Yes, I noticed that. I wasn't sure if it was safe to simply change the displayOrder or not. This is what I had in this PR originally, with ROR at the bottom (displayOrder=8):

However, I just went through an upgrade scenario and it looks like reloading a metadata block caused new displayOrder values to be saved without having an ill effect on older data. (I think I may have known this once but wasn't sure. Perhaps we should document it.)

Here's how I tested it. First, while running the "develop" branch, I added a couple authors with ORCID (first) and ISNI (second):

Then I edited citation.tsv to put ROR after ORCID as shown in the mockup. Since it worked (see below) I went ahead and pushed this change: 029c247

Then I reloaded citation.tsv:

curl http://localhost:8080/api/admin/datasetfield/load -H "Content-type: text/tab-separated-values" -X POST --upload-file citation.tsv

Now ROR appears second, right after ORCID:

I checked the database and (as expected) ROR has a higher ID than other Author Identifiers (since it was added later) but the sorting looks right, looks like what we want:

dataverse=# select * from controlledvocabularyvalue where datasetfieldtype_id = 11 order by displayorder;
  id  | displayorder | identifier |   strvalue   | datasetfieldtype_id 
------+--------------+------------+--------------+---------------------
   52 |            0 |            | ORCID        |                  11
 8569 |            1 |            | ROR          |                  11
   53 |            2 |            | ISNI         |                  11
   54 |            3 |            | LCNA         |                  11
   55 |            4 |            | VIAF         |                  11
   56 |            5 |            | GND          |                  11
   57 |            6 |            | DAI          |                  11
   58 |            7 |            | ResearcherID |                  11
   59 |            8 |            | ScopusID     |                  11
(9 rows)

So, I think we're good. We'll see what people say in review and QA. 😅

Again, I'm happy to document the behavior of reloading the metadata blocks to say you can reorder controlled vocabulary values. I was happily surprised. 😄