OK - There were actually two points hidden in my email.

The first (relationship of sampling and observations) was really just a throwaway. Probably shouldn't have even gone there in the context of this conversation.

But please note that sampling and observations are very very clearly separated in O&M (as is the feature-of-interest from an observation (the former is usually a sample), and as is the notion of 'result' from 'observation' (the latter being an action or event which generates the former)). These are the key features of the O&M model, the development of which was triggered by conversations where these things were confused. O&M was written up as an attempt to disentangle things (it is rooted in earlier "patterns" work from Fowler and O'Dea). Please don't surmise what might be said in O&M without taking a look - there is no conflation. (The perversity I referred to is rather that a model for sampling could be established without describing the relationship with observations, but I was being a smart-aleck and probably shouldn't have even mentioned it. In O&M there is merely a loose link between them, in that a SamplingFeature has an (optional) property 'relatedObservation' to link a sample to observations made on it. )

The second point is that, in the fields where I have experience, the key sampling metadata relates to provenance. Since there is now a standard ontology for provenance, maybe a sample model can be derived from this. Hence

sam:Specimen rdfs:subClassOf prov:Entity , sam:SamplingFeature .

Simon

----------------------------------------------------------------------

Message: 1 Date: Tue, 2 Sep 2014 08:41:50 -0600 From: Robert Guralnick Robert.Guralnick@colorado.edu Subject: Re: [tdwg-content] Darwin Core: proposed news terms for expressing sample data To: Simon.Cox@csiro.au Cc: TDWG Content Mailing List tdwg-content@lists.tdwg.org, Ramona Walls rlwalls2008@gmail.com Message-ID: CADAgxGWFmcKgEfYhfQ6u9=7gEzRgSSDdguaaOiPVcwK=ftwz=g@mail.gmail.com Content-Type: text/plain; charset="utf-8"

Simon --- Perhaps your email also illustrates Ramona's point about the complexity of the landscape right now when it comes to the _details_ of how we might model samples, and sampling processes. Darwin Core recently added a materialSample term, that was a refinement of OBI's "specimen" ( http://purl.obolibrary.org/obo/OBI_0100051) and with a description: "A resource describing the physical results of a sampling (or subsampling) event. In biological collections, the material sample is typically collected, and either preserved or destructively processed" ( http://rs.tdwg.org/dwc/dwctype/MaterialSample)

Maybe what you and what others are developing are equivalent expressions and perhaps they are not, but my initial reaction was that separating specimens and observations is a remarkably important and clarifying idea, and not perverse at all. I think the difference is that what you might call an observation is what the Biocollections Ontology community would refer to as a "sampling process" and what we call an "observation" is more related to some documentation of a "thing" in nature.

I say this recognizing that others, even those working on the Biocollections Ontology or Darwin Core, may have radically different notions of those same concepts. I hope not, but this is not simple to model. What I would like to see is some clarity among those putting in all this effort to do so more strategically, with bridges built across efforts and locations.

Best, Rob

On Mon, Sep 1, 2014 at 10:31 PM, Simon.Cox@csiro.au wrote:

Hi Ramona -

I understand your concern, though I would counterpoint that the only real reason to collect and curate a specimen is to support observations, either contemporaneously or at some future time. So it could be seen as slightly perverse to suggest that a model for specimens and samples could be divorced from the notion of observations.

FWIW I'm right now trying to develop a simplified SamplingFeatures ontology, still conceptually based on the ISO 19156 model, but with no commitments to marginal ontologies (i.e. lift it out of the ISO 19100 ghetto). This has led me to consider re-use of more standard ontologies. W3C Prov-O is interesting. Since a lot of the information that you would want to record about a specimen concerns its provenance, then it probably makes sense to align with prov. Currently I have

sam:Specimen a owl:Class ; rdfs:comment "A Specimen is a physical sample, obtained for observation(s) normally carried out ex-situ, sometimes in a laboratory."^^xsd:string ; rdfs:label "Specimen"@en ; rdfs:subClassOf prov:Entity , sam:SamplingFeature ; rdfs:subClassOf [ a owl:Restriction ; owl:cardinality "1"^^xsd:nonNegativeInteger ; owl:onProperty sam:samplingTime ] .

Simon