VIVO-1108: How should we model co-author affiliation when we only know it for one slice of time?

[chenejac]

Paul Albert (Migrated from VIVO-1108) said:

At some point (later this year probably), we will need to represent our affiliations of non-WCMC co-authors in VIVO.

Naturally, we do not have an authoritative source for this data. What we do have is an assertion from Scopus that at the time the paper was published, a given author had a reported affiliation.

This really complicates the standard way for representing people and appointments. I'm not sure we can assume that someone whose affiliation was the same in 2006 and 2008, had that same affiliation in 2007. Or, that a person who had affiliation A in 2008 ceased to have that affiliation if that person had a different affiliation in 2009. This person could have dual affiliations!

How can we assert affiliation without causing a mess?

Not open in Github at this time.

[chenejac]

Jon Corson-Rikert said:

From 5/15/13 Ontology call: That seemed to be a real stumper, because the fix version was delayed until post-1.6. At the time, I put in the ticket our publication ingest wasn't working and the Scopus API was outputting XML showing institutions in a big lump rather than tied to each author. Now though both of those roadblocks have been overcome. Also, unlike before, I have a not terrible approach to this. But I'm not sure if it's ideal.

For every author in the data, there is a Scopus-normalized institutional name (e.g., the author lists "Weill Medical College of Cornell University" and Scopus maps that to "Weill Cornell Medical College"), and a Scopus-generated institutional identifier (e.g., 60007997). In cases of multiple affiliations, you will see multiple institutional names and identifiers.

Here's the approach:

1. Create an object property for InformationResourceInAuthorship called ScopusInstitutionalAffiliationID.
2. ScopusInstitutionalAffiliationID has a range of foaf:organization.
3. Give foaf:organization has a data property of ScopusAffiliationName.

This will be really useful for us if we can bring these data in: it will populate our Dashboard, we can see how often our faculty actually list our institution, and we can identify patterns of inter-institutional collaboration.

Ensuing Discussion: there is so much variation in how people list their affiliation – will there be confidence in the identification of the organizations? Paul is primarily interested in querying the changes in affiliation for an author, both for disambiguation. Also wants to flag any document where there's no Weill Cornell affiliation, either because it predates when someone was working at Weill for some other reason.

will the VCard be ready for 1.6? some trickiness because the VCard spec is changing in a W3C and it's not clear even what namespace it will use. We at least know we want a structured set of strings linked to an authorship that can accommodate name parts, affiliation information, and potentially an institutional identifier.

[chenejac]

Jon Corson-Rikert said:

Brian -- can we consider this resolved by the VCard availability in 1.6? Or do we need to add UI controls to manage more than one VCard, as well as remove the assumption elsewhere in our code that every Authorship connects to a foaf:Person (or Agent).

Perhaps resolve and close this issue and create a new issue for 1.7 that states what you feel we need to do in that release to support Paul's intent.

[chenejac]

Jon Corson-Rikert said:

The VCards are the structure to hold these, but we need development to have additional listview support for showing secondary VCards.

Relates to holding archival information in VIVO as well.

[chenejac]

Paul Albert said:

This is not a front burner issue for us, but it may be worth tackling eventually.

In having some time to think about this, I would clarify that the point of this request was to have somewhere to capture the red rectangle in the attached. That would make links between institutional VIVOs easier.

There are different approaches here. You could do a string / data property like the below. You could input an institution in the form of linked data. Scopus has institution IDs. So does GRID.

Anyway, feel free to put this on hold if you like.

  !Screen Shot 2017-08-06 at 2.47.53 PM.png!

[chenejac]

Mike Conlon said:

Perhaps there is nothing to do here.  A co-author is a person.  A person can have a time limited position in an organization.  We have everything we need to model a person at an organization in a position for a known period of time.  No vcard is needed.