Open dshorthouse opened 4 years ago
The controlled vocabulary for role is a bit confusing. We've just added the labels "primary collector role" and "specimen collector role" in this field (see https://www.gbif-uat.org/dataset/038da6db-3838-44fe-b58d-6a43f3f4bb31). Would it make sense to have an additional roleID field for the CRO_ numbers identifiers?
I'm also not really clear why the "primary collector role" and "specimen collector role" are necessary at all. Surely they are covered by the displayOrder, and one could just have "collector"?
It also strikes me that there's a big overlap with action here; all of the actions listed (e.g. georeferenced) imply the roles pretty heavily - what other role could there be than "georeferencer"? Perhaps it doesn't matter if there's some redundancy here though?
These are all good points and questions. The use of role
here is an attempt to align with the work done in the context of the Research Data Alliance. See http://doi.org/10.5334/dsj-2019-054. The intent of this is to express some granularity or differentiation among agents that were involved in the execution of the same action on a particular item. A value for role
is indeed peculiar when there is a single agent executing the action - there's no ambiguity whatsoever. Likewise, the order of agents to a degree accommodates the spirit of what role
is meant to represent (i.e. Mary is listed first among co-collectors so she must have played a different role than all others listed after her). What we're trying to do here is point to a single agent among many involved in a single activity, regardless of their ordering, and give some precision. In Botany, it's the primary collector that has their specimen collector number on the sheet (= recordNumber
). But, outside this particular example, I grasp at coming up with terms for role
that fit other activities than collecting. @diatomsRcool any thoughts here?
For a georeferenced occurrence, you might have one person with role georeferencer
and another person with role quality checker
, I think that would be valid right?
The use of role is controversial. In the context of RDA, we had strong feelings for and against. We knew we needed to be able to accommodate role while also not requiring it. Thus, we build out the role vocabularies for the people who need them. The above example is a good one. Would the quality assurance role be a subclass of a curation role?
Would the role
of georeferencer
apply to the person in relation to an Occurrence instance? Or more properly to a Location instance associated with that Occurrence instance? I know most of the discussion in this attribution context is focused on the relationship between agents and occurrence instances, but agents play important roles in all of the DwC classes, as well as highly-relevant instances of things not represented in DwC classes (e.g., References).
Regardless of whether there is a distinction between role
and AgentAction
, there ought to be clarity on not just the Agent performing the action, but the specific object on which the action was performed.
role (property)