search

mdoering / dwca-examples

Documenting Darwin Core Archives By Examples
4 stars 0 forks source link

Terms to indicate the type of record #2

Open peterdesmet opened 10 years ago

peterdesmet commented 10 years ago

There are currently 3 properties to indicate the type of record: rowType, dc:type and dwc:basisOfRecord. These properties can only be used in certain combinations (so there seems to be a relationship between those) and the latter two are controlled with a vocabulary.

The summary of use document proposes two possible new type properties: eventType and materialSampleType. Before we do this, I think it is worthwhile to clarify the current values and their relationships. I also wonder if some the listed values in dc:type and dwc:basisOfRecord are used at all, and if it is necessary to add new type properties.

rowType

Seems to be based on a Darwin Core class for cores:

Occurrence
Taxon
Event
MaterialSample

Seems to be based on a Darwin Core class or the name of the extension for extensions:

MeasurementsOrFacts (class)
Occurrence (class)
Multimedia
TypesAndSpecimen
Distribution
Multimedia
VernacularName

dc:type

Controlled vocabulary defined in http://dublincore.org/documents/dcmi-terms/#H7

Collection
Dataset
Event
Image
InteractiveResource
MovingImage
PhysicalObject
Service
Software
Sound
StillImage
Text

Are all these terms currently used by our community?

Note that Event is currently used for observations (rowType:occurrence) and is planned to be used for Events (rowType:Event). Won't this be confusing?

dwc:basisOfRecord

Controlled vocabulary defined in https://code.google.com/p/darwincore/wiki/RecordLevelTerms#basisOfRecord

Occurrence
Event
Location
Taxon
PreservedSpecimen
FossilSpecimen
LivingSpecimen
HumanObservation
MachineObservation
NomenclaturalChecklist

Are Event, Location, and NomenclaturalChecklist currently used by our community?

Taxon seems useful, but the Taxon Core does not have a basisOfRecord (nor type), so how can it be used?

Relationships

Here's an attempt:

rowType dc:type dwc:basisOfRecord What
Occurrence PhysicalObject PreservedSpecimen, FossilSpecimen, LivingSpecimen Specimens
Occurrence Event HumanObservation, MachineObservation Observations without proof
Occurrence Image, MovingImage, StillImage, Sound HumanObservation, MachineObservation Observations with proof
Occurrence Event Occurrence Unspecified occurrence
Taxon field not available field not available
Event ? ?
MaterialSample ? ?

I find the combination of these rather complex and I think that is also the case for data publishers. I'm especially wondering if dc:type is useful at all.

mdoering commented 10 years ago

Thanks for putting this together Peter. I think we really need to think this through from scratch and come up with something simple that satisfies what people really want to know. It might be all it is is the "kind of proof" property? For occurrences there was a proposal once in TDWG to have an evidenceType term instead of BoR and dc:type that simply indicates the kind of proof there is. If that terms would allow multiple values coming from a hierarchical vocabulary with things like Physical (Fossil, Preserved, Sample, Living, Sequence), Media (Image, Video, Sound, Literature) it seems we would be in a better shape. If I wanna know what kind of sample (or event) has been taken I think I would prefer to have a separate vocabulary for those and a new term like materialSampleType or eventType