New dataset resource

[cgendreau]

The concept of a Dataset in DINA is yet to be clearly defined.

With the advancement in DwC-A export functionality a source of data to produce an EML file is necessary.

Dataset

• Multilingual Title and Description
• Dataset type (DwC-A, DataPackage)
• Agents with roles
• Group (to control permission)
• Saved search used to produce the dataset (so we can generate a new one in the future)
• isPublic (the idea would be to know if we need to apply the isPubliclyReleasable filter)
• Version
• link to most recent version (ideally stored in the object-store)

[dshorthouse]

Great start! Other fields, some evident some not so much, loosely based on EML:

• data license
• managed attributes
• timestamp when created (associated with version)
• external identifiers (for purposes of linking out to where registered/shared, e.g. GBIF)
• anticipated update frequency
• new functionality: auto-generated to coincide with update frequency
• acknowledgments
• citations

[jgrieb]

Are we discussing here adding Dataset as a core entity to the collection specs? If yes I do see some semantic overlap with e.g. Collections or Assemblages, and surely often it will be a use case to produce a DwC archive of all samples belonging to Collection X/ Assemblage Y....

Or will this be more an "implicit" entity, e.g. a form which a user fills out when requesting the DwC export, and potentially the values are cached somewhere in the user preferences...?

[cgendreau]

This would be a new core entity.

There are similarity with collection and assemblage but a dataset can always have a mixture of everything depending on how it is structured. Research dataset can have material sample from different collections and they may or may not be part of an assemblage.

For simple cases the dataset will simply be a query to get everything from a collection.

[dshorthouse]

@jgrieb I had assumed it would be similar (in spirit) to a "Project", what I'd call administrative metadata, except insofar as having any explicitly specified hard links to any object. So...equivalent to an EML form as would be in an IPT but with some additional functionality to craft a query for what gets flattened/exported and at what frequency. Future challenge(s) would be what to do about extensions like identification history or multimedia if we were to envision a fully fleshed-out DwC-A. As a first pass, assemblage associations could be a queryable item for use as a filter in an "occurrence" core export as could any number of other criteria on material samples such as "publicly releasable" flags.