How to model a 'ratio' using the P01 semantic model? (BODCNVS-1789) about p01

[roy-lowry]

Roy is around but now only looking at e-mail about once a week.

[meganrwong]

Here are some other examples I’m unsure how to model. These parameters involve calculations (sums, ratios, indices/other calculation) with more than one chemical entity

The parameter is 'concentration of ammonium-N and nitrite-N (NH4 + NO2)-N'

1. What would the property (S06) be?
2. What would go in the chemical entity (S27)?

Parameter ‘anion storage capacity’
3. What would the property (S06) be?
4. What would go in the chemical entity (S27)?

and there are more examples ......
P buffer index (PBI)
phosphorus buffering capacity (PBC)
P saturation ratio
Colwell-P/PBI (ColP) ratio (ratio of two soil P results)
Cation exchange capacity (CEC) and Exchangeable Cation Exchange Capacity (ECEC) – (depending on the method, these re different calculations using elements Ca2+/Mg2+, Na+, K+ - calculations specified as part of method)
Base saturation percentage (BSP)
Ratio of calcium to magnesium (Ca2+/Mg2+) (a ‘basic cation saturation ratio’)

[gwemon]

Over in nvs-vocabs/S06#74 (comment) we have a parameter that we are unsure how to model using the semantic model.

The parameter is 'ratio of X/Y in the sample'.

1. what is the S06 'property'
2. how do we link chemicals X and Y into the parameter definition?

@roy-lowry surely you have seen this before?

Hi @dr-shorthair Sorry for slow response I was on leave and Roy must be away too. We have 3 types of vocabularies to define the "object of interest" of a measurement variable: one for when the object of interest is a biological entity (S25), one when it is a chemical entity (S27) and one when it is a physical entity or anything that is neither a chemical nor a biological entity. The latter is the S29 vocab. This is where we decided to define ratios of chemical substances. See examples such as: http://vocab.nerc.ac.uk/collection/S29/current/PE001382/
As you will see we used the SKOS core#related property to link the S29 entity to its S27 components.

[roy-lowry]

@dr-shorthair There are existing examples of chemical ratios such as CNWCXX10 that are either 'Atomic ratio' or 'Mass ratio'. Couldn't find any examples of 'Concentration ratio' though. You could have a look at the modelling for these for a precedent.

@meganrwong There are many examples for summed concentrations of multiple chemical species such as NTRZMOD1, which (ignoring the 'N' issue) is similar to your example.

[gwemon]

@meganrwong I was going to answer and I see that Roy has just done so.

[dr-shorthair]

Thx @gwemon and @roy-lowry - that clarifies a lot. Anything that isn't a chemical or biota is a physical entity (including ratios).
I now see that some, though not all, of the S29 chemical ratios have explicit related links to their component concepts from S27 - compare http://vocab.nerc.ac.uk/collection/S29/current/PE001378/ with http://vocab.nerc.ac.uk/collection/S29/current/PE001391/.

However, it is not clear how this relates to the Semantic Model.

As you know, I have implemented the Semantic Model as an OWL/RDFS ontology - see https://github.com/CSIRO-enviro-informatics/PUV-ont/blob/main/README.md.
In particularly, see the summary https://raw.githubusercontent.com/CSIRO-enviro-informatics/PUV-ont/main/image/puv-Parameter.png
This provides semantic labels to the relationships between the elements of the parameter definition.
Looking at the way you are composing a Physical Entity for a ratio of Chemical Entities, and a sum of entities, it appears that a couple of relationships are missing from the Semantic Model, at least as it is shown in the summary diagram.

1. Is the summary diagram complete, or are there some additional relationships that should be incorporated in the RDF implementation?
2. In particular

• if Chemical Substance (S27) was also a sub-class of Physical Entity Component, then a Physical Entity could refer refer to the entities to build ratios, etc
• if there was an is composed of link from Chemical Entity back to itself, this would allow for entities that are composed of multiple chemical species for summed concentrations etc.

Finally,
    3. in our application, there are a variety of 'qualifiers' of chemicals used, including soluble, exchangable, pseudo, oxidisable, extractable, reducible, titratable. Do you think these should be included in separate S27 entities, in the same way that you have Organic carbon and easily-reducible total iron

[meganrwong]

Thanks everyone. I appreciate your very helpful responses @roy-lowry and @gwemon
Simon has covered all of the follow up questions we have
I am following with interest :) Thanks again

[gwemon]

Thanks @meganrwong @dr-shorthair Indeed it would be good to add this to the semantic model. We also need to consider this in the framework of the latest developments with the I-ADOPT model. Maybe we could have a short meeting so that we can put your work in context with what we are currently working on?

[meganrwong]

Thanks for all the thought folk, this has been giving the model and its application a good workout. I've had to make a decision re our soil chem so logged it here -
#ANZSoilData/def-au-parameters/issues/6

Also, discussion prompted this issue #ANZSoilData/def-au-parameters/issues/5

[meganrwong]

also added the chem qualifiers question over at repo for soil chem parameters for use in Aust National Soil Info System (ANSIS)

[meganrwong]

Thanks @meganrwong @dr-shorthair Indeed it would be good to add this to the semantic model. We also need to consider this in the framework of the latest developments with the I-ADOPT model. Maybe we could have a short meeting so that we can put your work in context with what we are currently working on?

Hi Gwen - noted :) - Simon was reflecting on pts 1 and 2 above

[dr-shorthair]

How about this:

Note: puv:ChemicalEntity == "Chemical Substance (S27)"

In which case we could see

<http://vocab.nerc.ac.uk/collection/S29/current/PE001378/> a puv:PhysicalEntity ;
    skos:prefLabel "neodymium-143/neodymium-144" ;
    [...]
    puv:isComposedOf <http://vocab.nerc.ac.uk/collection/S27/current/CS002797/> , 
                  <http://vocab.nerc.ac.uk/collection/S27/current/CS003788/> ;
.

and

<http://vocab.nerc.ac.uk/collection/S27/current/CS026905/> a puv:ChemicalEntity ;
    skos:prefLabel "nitrate+nitrite" ;
    [...]
    puv:isComposedOf <http://vocab.nerc.ac.uk/collection/S27/current/CS002879/> , 
                  <http://vocab.nerc.ac.uk/collection/S27/current/CS002877/> ;
.

Using the same predicate all round still doesn't reflect the maths, but does provide explicit links to the component chemicals, which would help with some functions (e.g. discovery).