Skip to content

Latest commit

 

History

History
190 lines (131 loc) · 6.15 KB

2021-11-09.srap_meeting.md

File metadata and controls

190 lines (131 loc) · 6.15 KB
Raw

DC SRAP meeting 10

Time: 2021-11-09, 16.00-17.00 CET (15:00-16:00 UTC)

Present: Osma (chair), Karen, Tom (minutes), Jan, Nishad, Alasdair

These minutes: https://github.com/dcmi/dc-srap/blob/main/meetings/2021-11-09.srap_meeting.md

  1. Opening of the meeting
  1. Appointment of the minutes taker (Osma will act as the chair)
  1. Approval of the agenda

Osma: Other issues for the agenda?

Tom: Juha sees a dependency on something like the Porto mechanism for encoding persistent identifiers in XML attributes, with regard to Affiliation.

Alasdair: Plan to reactivate working group.

Osma: Agenda is approved.

  1. Minutes of the previous meeting (2021-10-18)
    • draft minutes were posted to the dc-srap list by Alasdair
    • finalized minutes to be put on GitHub

Osma: Minutes of previous meeting approved.

Alasdair will upload.

  1. Element-specific discussion

Publication status (discussed at the previous meeting)

  • New Date subproperties (#20)

Karen: Various concerns: We need good definitions because they are relations between two things. Replaced, retracted and missing - Juha wanted to add. But not clear what record this would be added. DCMIMT has isReplacedBy/replaces which are relations. If you use isReplacedBy and link to other document (desirable), then you only need a date, not a "date replaced by".

Osma: In what setting is this important?

Karen: Replaced and retracted are important in scholarly work. "Something has changed that matters." And I have another problem with "missing" - can return to this. Juha wanted to take this into vocabulary but we need good definitions.

Osma: In case of "retracted" (and "missing" and "lost"), no new record.

Karen: Issue with date retracted is that you should have only the metadata - not the item.

Osma: What is the way forward?

Karen: Need to get definitions, but we only have the terms. So would help to have Juha's definitions. Cannot accept the terms without definitions. I do not have definitions.

One other issue: Not sure what "missing" means. If it means you cannot find it, it is not a publication status, but an "archive" or "library" status. Juha cited example of whole journals disappearing. If publisher disappears, one says "defunct". "Missing": publication status, or a holdings status?

Jan: More like "holdings" statement.

Karen: Will open a new Github issue, try to flesh out.

Jan: DataCite has "obsolete" - denotes that something is obsolete, which is a publication status.

Osma: To me, obsolete is a kind of replacement.

Affiliation #3

Osma: Had some discussion on Github issue. Juha's original proposal. Question of how to provide persistent ID and name in XML metadata. Uses syntax drafted around Porto meeting. But also more questions about how to express affiliation.

Karen: Wasn't aware of Porto work so referred to fact that terms need to refer to the proper resource - eg, affiliation as property of person.

Osma shows original proposal for SRAP application profile. Diagram shows affiliation as a property of the Agent. Dates when an affiliation is valid. Wikidata has ways to express qualification.

Karen: So if affiliation is an organization...

Tom: The PID discussed in Porto was about PIDs for the person itself rather than affiliation per se.

Nishad: Could be easy to follow model of ORCID.

Osma: Similar to Wikidata model. If we want to support that, would need, say, Affiliation class.

Nishad: Current publication model will state the affiliation as a point in time. Potential: update this information from ORCID. Affiliation should have a qualifier.

Osma: What affiliation at the time paper was written?

Nishad: I can go to my ORCID page and disable affiliations I do not want to show, update information.

Jan: Need both - affiliations then, and up-to-date contact data (can get from ORCID).

Osma: Dspace pointing out to system that provides this information. Dspace has grown with respect to the contextual information it can store.

Two things need to be sorted out:

  • What is the model for expressing affiliation? Original model seems too simple.
  • We need some solution for providing IDs in attributes. Hopefully, not using "id=" because this has special meaning in XML.

Karen: What domain classes?

Osma: Picturing:

  • Agent and Organization, but need box in the middle, Affiliation, which would connect the two and have other properties: role, source.

Tom: Complexify the model? Or consider out of scope and assume the profile will point to something with that detail?

Osma: This has been driven by requirements of existing profile used in Finland. Could be considered out of scope? Invent another way to describe Agents?

Karen: FOAF?

Osma: Wouldn't give us a way to express the start and end dates.

Karen: But would give us the enough for the simplest scenario. Would like to see a simple solution, done as Dublin Core or FOAF.

Osma: Should think it would be easier to come up with example documents. "If we do it like this, will look like that". Defining domains and ranges would take more effort.

Karen: I find the specification easier.

Osma/Karen: Examples must come first (agreed).

ACTION: Osma will draft some examples, simple as possible, for affiliation, then more complex, like Wikibase or ORCID.

Karen: Let's start with those two. Then I will do in FOAF.

isVersionOf and hasVersion

  1. Any other business
  1. Next meeting
  • Confirmed: 2021-11-23, 15-16 UTC (2 weeks from now)
  1. Closure of the meeting