Open alexrobin opened 5 months ago
Yes - this was intentional. The mismatch was discussed here: https://github.com/w3c/sdw-sosa-ssn/issues/51#issuecomment-1738334326 etc.
The choices of new classes were based on articulated requirements. No-one asked for SamplingCollection and it is less obvious that it would be used. As you know, ActuationCollection was only introduced recently in response to your use-cases.
I recognise that there is a gap that symmetry suggests could be filled. OTOH it could also be argued that Sampling is a subclass of Actuation, so perhaps that conversation should be had first.
@dr-shorthair
I've come across this same issue, I need to use a SamplingCollection
to group a few hundred thousand Sampling
instances with common properties (same usedProcedure
, same hasFeatureOfInterest
, etc).
As Observation
and Sampling
are both actions, the use case would be exactly the same as how ObservationCollection
is used, simply reducing the number of triples in a Graph by hoisting common properties to the collection level. I agree with the final comments in #51, it seems like when creating SampleCollection
there was a missed opportunity because SampleCollection
is at the level of ResultCollection
, and SamplingCollection
should have been created as sample-counterpoint to ObservationCollection
.
In the absence of SamplingCollection
, what is the recommend method to achieve this result?
No-one asked for SamplingCollection and it is less obvious that it would be used.
Now it looks like @ashleysommer is asking. Good. (I hope he can also provide 'evidence of implementation' in support ;-) )
To start the process of also adding SamplingCollection
please branch/fork and make a PR.
At a minimum, the following files will need changing:
@ashleysommer could you prepare a PR to address this?
Thanks for your response @dr-shorthair Yes, I definitely intend to submit a PR for this, I just need to find the time this week to get it written up.
In regard to 'evidence of implementation', what does that look like? Is it an example usecase of how we'd like to apply SamplingCollection
in our datasets, or would it be something like a PR to include the SamplingCollection
profiled in the TERN Observations Ontology?
The latter would suffice.
@dr-shorthair
~I notice as part of the introduction of SampleCollection
it was decided it could include properties of both Sample
and Sampling
, that seems a bit odd to me (notably hasSampledFeature
and hasOriginalSample
) see attached image.~
~Should these be removed from SampleCollection
with the introduction of SamplingCollection
as the intention is SamplingCollection
will capture properties from Sampling
?~
EDIT: Actually, perhaps I got it around the wrong way, that image is incorrect because it doesn't match the domainIncludes
of hasSampledFeature
and hasOriginalSample
in the ontology, or in the Sampling.html
sosa:hasOriginalSample
a owl:ObjectProperty ;
rdfs:label "has original sample"@en ;
skos:definition """
Link to the original sample collected from the ultimate feature of interest that is related to the context Sample or SampleCollection through a chain of isSampleOf relations.
"""@en ;
schema:domainIncludes sosa:Sample ;
schema:domainIncludes sosa:SampleCollection ;
schema:rangeIncludes sosa:Sample ;
rdfs:isDefinedBy sosa-sam: .
sosa:hasSampledFeature
a owl:ObjectProperty ;
rdfs:label "has ultimate sampled feature"@en ;
skos:definition """
Link to the ultimate feature of interest of the context Sample or SampleCollection - i.e. the end of a chain of isSampleOf relations
"""@en ;
schema:domainIncludes sosa:Sample ;
schema:domainIncludes sosa:SampleCollection ;
schema:rangeIncludes sosa:FeatureOfInterest ;
rdfs:isDefinedBy sosa-sam: .
Yes - you are correct. The diagram is wrong. The original proposal was here https://www.w3.org/TR/vocab-ssn-ext/#ultimate-feature-of-interest Figure 2. I think this would be the correct diagram?
Yep, that revised diagram looks correct to me.
Where is the source file for that diagram? It would be good to be able to add SamplingCollection
to it for the PR.
I sent you a link. Suggest adding SamplingCollection in upper right, in common with Actuation and Observation diagrams.
Okay, I got the link, thanks Simon.
It looks like both the ObservationCollection
and ActuationCollection
are on the upper left of their respective diagrams, so I've done the same for SamplingCollection
. How does this look?:
You are right. Mild dyslexia strikes again
Here's the work-in-progress Collections diagram, will be submitted with the PR:
Another thing I noticed, it looks like in the Observations diagram and the Actuation diagram (and the Collections diagram), ObservationCollection has hasResult
but not hasSimpleResult
, and ActuationCollection
has neither hasResult
nor hasSimpleResult
but in the ontology files both collections have hasResult
and hasSimpleResult
.
@ashleysommer would you be available to give a briefing on your proposal at a Wednesday evening telecon? People are interested to hear your use-case.
I noticed that we have added
ObservationCollection
,ActuationCollection
andSampleCollection
to SOSA.But the actual equivalent of
ObservationCollection
andActuationCollection
in the sampling module would beSamplingCollection
.While
SampleCollection
is also useful, I was wondering if the omission ofSamplingCollection
was intentional.