search

tdwg / dwc-qa

Public question and answer site for discussions about Darwin Core
Apache License 2.0
49 stars 8 forks source link

Defining indexes calculated in a post-survey analysis #148

Open jennet opened 4 years ago

jennet commented 4 years ago

Does Darwin Core (I'm interested in the Darwin Core RDF) have a best practise way to describe values that are calculated post-survey during an analysis of the data?

e.g. I have a sample date, method and then an analysis date and method - together with values for indexes such as the BMWP for invertebrates.

Does Darwin Core cover this type of information? or does it restrict itself to occurrence counts and not derived/calculated data

baskaufs commented 4 years ago

This is not a particularly well-developed area as far as RDF is concerned.

The RDF Guide recommends literal values for dwc:measurementDeterminedDate, dwc:measurementValue, dwc:measurementAccuracy, and dwc:measurementRemarks with controlled value IRIs for dwciri:measurementType, dwciri:measurementUnit, and dwciri:measurementMethod (although those controlled vocabularies aren't specified). But those terms are really for measured values rather than calculated ones.

So I guess I'd say that Darwin Core itself doesn't really deal with post-survey data.

The Darwin-SW model (DSW), which is not an official part of any TDWG standard, does offer a way to model derived resources through the transitive term dsw:derivedFrom. One could assert that the measurement was dsw:derivedFrom an organism or organisms and that the calculation was dsw:derivedFrom the measurement. That would be a way in RDF to establish a link between the calculated values and ultimately the organism(s) they were derived from. Because DSW was designed to work with Darwin Core, the derived things (which can be of any class) are shown connected to dwc:Organism class instances. However, dsw:derivedFrom doesn't have any defined range, so it would be fair game to use it to connect pretty much any class instances.

I don't have any idea how you would model the measurement data and calculated data instances. I'm sure somebody has worked on that, but I'm not up on that subject. Basically, I think you are in uncharted territory! Good luck!

jennet commented 4 years ago

This is great info, thanks. dsw:derivedFrom looks very interested, I'll look into that