Clarify cdt:ucum and QUDT explanations

[dr-shorthair]

The pattern for scaled values https://w3c.github.io/sdw-sosa-ssn/ssn/#ModelQuantity shows examples from QUDT and UCUM.
Need to note the use of qudt:hasUnit on execution collections mixed with sosa:hasSimpleResult on members does NOT give license to mix qudt:hasUnit on individual executions outside of qudt:QuantityValue structures.

Also, need to add verbiage in https://w3c.github.io/sdw-sosa-ssn/ssn/#SOSAhasSimpleResult with some cautions about custom datatypes (including cdt:ucum and GeoSPARQL's geo:wktLiteral) which may cause some RDF/OWL processors to barf.

[dr-shorthair]

@oldskeptic does this description capture the main concerns?

[maximelefrancois86]

It does.

Note we haven't talked about the fact that the input value can also be expressed as a quantity value. As a consequence, it seems uneven that the qudt:hasUnit property on an Execution collection only propagates down to the result of individual executions. One could add that it also propagates to the input values.

In addition, one could suggest that qudt:hasUnit can also be asserted at the Execution level, meaning it would propagate down to any input value or result of that execution

[dr-shorthair]

Well spotted. Yes, we need comment on the scope of qudt:hasUnit when present on a Collection.

If the standard consistency rules were applied, then all member Executions would be constrained to use the value given on the Collection.
However, arguably the 'standard' consistency rules don't apply since

(i) qudt:hasUnit is not part of SSN
(ii) qudt:hasUnit is not a property of the member Observations, it is a property of one-or-more of the (RDF) properties of the Observations

What kind of rule should we suggest applies in this case, do you think?

[minorg]

I think the spec should be more opinionated about preferring n-ary forms with QUDT over custom datatypes, given the dearth of implementation support for cdt:ucum in SPARQL engines. The literal-with-units form looks nice to the human reader, but there are good reasons that approach hasn't caught on, despite multiple people trying it over the course of years.

[dr-shorthair]

Thanks @minorg
Could you propose a specific change? Best of all as a branch/PR

This is the primary file

• https://github.com/w3c/sdw-sosa-ssn/blob/gh-pages/ssn/chapters/ModelQuantity.html

But cdt:ucum also appears in

• https://github.com/w3c/sdw-sosa-ssn/blob/gh-pages/ssn/chapters/Common.html
• https://github.com/w3c/sdw-sosa-ssn/blob/gh-pages/ssn/chapters/ModelDomainTypes.html

as well as several files in https://github.com/w3c/sdw-sosa-ssn/blob/gh-pages/ssn/rdf/examples

• bubble-provenance.ttl
• IBS-TH2-PLUS.ttl
• IDEAS.ttl
• ip68.ttl
• iphone_barometer-sosa.ttl
• iphone_barometer-sosa.ttl
• obs-sample-foi.ttl
• paleo-atmosphere.ttl
• spinning-cups.ttl
• Thermometer.ttl
• timeseries-oc.ttl
• UOM-cdt.ttl

[dr-shorthair]

Also see results of tests by @oldskeptic reported here: #382 (comment)