Add IdentificationProtocol table to DwC-DP

[tucotuco]

This proposal arises from discussion in tdwg/dwc#774, but requires a use case for multiple Protocols for a given Identification.

[CecSve]

A use case could be when a specimen in a collection is identified (potentially several times) by morphology, following various keys, and also being DNA barcoded for a reference DB. In this case, all used protocols would lead to the same identification, but there could, of course, be conflicting identifications coming out of it in a real-life scenario.

[tucotuco]

@CecSve Aren't keys references rather than protocols? They might all lead to the same conclusion, but through different processes. The conclusion isn't the Identification, the Identification is the combination of all the method, references, actors and conclusions, so each of the several times you are talking about seem to be different Identifications. That we can already support without an IdentificationProtocol table.

[CecSve]

Keys may be references, rather than protocols.

The only other example I can come up with is for DNA data, and that is well catered for in nucleotide tables and the molecular protocol table.

@ManonGros @kkuczkowska I may need your help on use case justification. As @tucotuco mentions, the printed key can be shared in Identification Reference with further info in the Bibliographic Resource.

[dshorthouse]

Could a named algorithm (or suite of them) used by an AI agent in the course of applying a scientific name to an organism represented in a photograph be considered an identificationProtocol?

Others: dissections, staining, serological techniques, biochemical tests, mass spectrometry

[timrobertson100]

I don't know if this is helpful but the identification section within biosecurity protocols may bring suitable case studies, such as
this example. Like taxonomic keys, I would think this would be a reference to link to, as I can't see the methodology translating easily to a tabular table.

[tucotuco]

Could a named algorithm (or suite of them) used by an AI agent in the course of applying a scientific name to an organism represented in a photograph be considered an identificationProtocol?

Others: dissections, staining, serological techniques, biochemical tests, mass spectrometry

Interesting. This makes me think that, in addition to an identificationID, we should add an identificationProtocols field that can hold string literals (including a list). This would parallel assertionProtocolID and assertionProtocols.