property:agentType

[dshorthouse]

agentType (property)

Definition	The nature of the agent. Recommended practice is to use a controlled vocabulary.
Existing property
Existing namespace
Existing property identifier
Format	string
Required	yes
Constraints	Controlled vocabulary
Examples	"Person", "Organization", "SoftwareApplication"
Notes	Thesaurus at https://tdwg.github.io/attribution/people/dwc/vocubulary/agentType.xml

[dshorthouse]

Though not explicit, the other possibility is a collecting Party, Team, or an Expedition. However, we do not have anything in schema.org for any of these scenarios. And, if higher-order groupings of people are used, we do not have a solution to express an individual's membership in such groups while simultaneously expressing the relationship to an occurrence as required in a DwC-A extension.

[PietrH]

How about placing collecting parties, and teams under something like programMembership as a child of a Organisation, with the inverse property memberOf and the property member for individuals?

Since Expeditions have a spatial and temporal component that is more defined than a Team, Group or Party, I would tend to see these as Events, and a specimen can be the subjectOf this event, with individuals being of type contributor or attending to this event? Would I be wrong to place an Expedition under a sampling event in DwC?

[wouteraddink]

I would rename this to agentType

[wouteraddink]

And likewise for all other properties that are about the Agent and not about the Action.

[dshorthouse]

@wouteraddink If we rename to agentType, we're making a new term. Do you know of an existing property & identifier that can be reused here?

[dshorthouse]

@PietrH wrote:

How about placing collecting parties, and teams under something like programMembership as a child of a Organisation, with the inverse property memberOf and the property member for individuals?

Would love to represent such a hierarchy, but we're constrained in the present exercise by what we can do in the star schema of Darwin Core with its single level of joins to tabular data. Do you think it's possible to flatten-out what you suggest as columns w/ terms?

[matdillen]

Why does this property reference dc:type, yet recommend the schema.org type vocabulary and not the DCMI one? I wouldn't use the term type if it doesn't use the Dublin Core vocabulary, as this is how it's done elsewhere in Darwin Core.

I think there may be some sort of precedent for Wouter's suggestion of agentType in gbif:typeDesignationType, dwc:measurementType and GBIF's internal use of mediaType. Although the latter does make use of DCMI's vocabulary and not another one.

[dshorthouse]

Why does this property reference dc:type, yet recommend the schema.org type vocabulary and not the DCMI one? I wouldn't use the term type if it doesn't use the Dublin Core vocabulary, as this is how it's done elsewhere in Darwin Core.

We did once use schema.org's type but it was recommended via email communications (@timrobertson100 I think) that we use dc:type. The examples had not yet been changed. However, dc:type in Darwin Core recommends the use of terms in a controlled vocabular, https://www.dublincore.org/specifications/dublin-core/dcmi-type-vocabulary/, none of which are what we need here.

I think there may be some sort of precedent for Wouter's suggestion of agentType in gbif:typeDesignationType, dwc:measurementType and GBIF's internal use of mediaType. Although the latter does make use of DCMI's vocabulary and not another one.

So, should we revert back to schema.org (what I'd prefer) or do we need a new term, agentType?

[matdillen]

I'd suggest changing to agentType and using schema.org as a vocabulary. This is still not ideal, but I think we should definitely avoid minting a term type which uses another vocabulary than the same term in other Darwin Core modules (core or extension).

[matdillen]

Should the vocabulary be documented here similarly to the other three properties with a vocabulary? This currently isn't the case for agentType.

[dshorthouse]

Because there are only three terms (one of which is a "maybe"?), I was reluctant to create a vocabulary for it. But, I suppose if we want to maintain a strict representation of this term & make it verifiable by machines, we ought to have such a vocab.

[matdillen]

This came up when I wanted to reference the vocabularies and noted there was one for role but not agentType. Whereas this is a required term and, by setting a vocabulary, we can already try to preempt some potential problems - like whether the full schema URI needs to be used or only the term itself.

By setting up a concrete vocabulary, the discussion on what should be a part of it (like SoftwareApplication) also becomes more actual.

[dshorthouse]

Vocab created just now.