Make MeasuredVariable a complex property

[RKrahl]

Change property MeasuredVariable, which is a simple free text property by now, into a complex property having the subproperties measuredVariableName, measuredVariableIdentifier and measuredVariableIdentifierType. This is done to allow linking with a term from controlled vocabulary. If that (external) controlled vocabulary defines persistent identifiers for its terms, that identifier would be stored in measuredVariableIdentifier.

This complex property follows the same structure that we also have for instance in Model or InstrumentType. Close #68.

The new structure of the property will be:

ID	Property	Obligation	Occ	Definition	Allowed values, constraints, remarks
10	MeasuredVariable	R	0-n	The variable(s) that this instrument measures or observes
10.1	measuredVariableName	R	1	Full name of the variable	Free text
10.2	measuredVariableIdentifier	O	0-1	Identifier used to identify the variable	Free text, should be a globally unique identifier
10.2.1	measuredVariableIdentifierType	O	1	Type of the identifier	Free text

[strambc]

• Create a new field called "MeasuredQuantity" and indicate that "MeasuredVariable" is deprecated.
• Give specific instructions on mapping between versions in documentation.
• Suggest ontologies in the documentation, as example values for measuredVariableIdentifer.

[paulmillar]

One constructive comment against @strambc suggestion:

• Update documentation to recommend publishing both MeasuredVariable and MeasuredQuantityName with the same value where backwards compatibility with older clients is important.

[schmidt4earth]

Points 10.2 (measuredVariableIdentifier) and 10.2.1 (measuredVariableIdentifierType) seem to me to me not to add much additional benefit. If implemented it should have examples to choose from.

[paulmillar]

Points 10.2 (measuredVariableIdentifier) and 10.2.1 (measuredVariableIdentifierType) seem to me to me not to add much additional benefit. If implemented it should have examples to choose from.

From my understanding, one possible value for measuredVariableIdentifier could be values from the "Quantity Kinds" Vocabulary from QUDT.

Here are some concrete examples (from QUDT):
- http://qudt.org/vocab/quantitykind/RelativeHumidity
- http://qudt.org/vocab/quantitykind/Activity
- http://qudt.org/vocab/quantitykind/TotalPressure

These URLs are both persistent identifiers (IRIs) of the term and something that you can put in a web-browser and see something meaningful for a person.

In general, by linking to such vocabularies, software that consumes PIDINST information can show related information: the measured quantity name in different languages, the symbol used, different units that are appropriate, etc. It also avoids ambiguity (spelling, terminology).

To be clear, I'm not advocating for using QUDT, it's just an example.

[RKrahl]

Discussed in today's meeting:

• there are concerns that this is an incompatible change.
• one option to allow for a smoother transitions between schema versions would be to rename the property: e.g. add the complex property as a new one having a new name as suggested in the comments above and retaining the old MeasuredVariable for some transition period, but deprecating it.
• postpone the final decision to the June monthly meeting to allow for some more time to discuss the details here.

[RKrahl]

Please vote!

We need your feedback in order to decide whether to add this property. Please vote by clicking on the emoji at the bottom of this comment and select either:

👍 -- accept: add this property
👎 -- reject: do not add it
👀 -- abstain

Please add your vote by end of Juin. We will discuss the outcome of the voting in the July monthly meeting. Please bear in mind that we have two alternative proposals (this PR and #90) for this issue and we will only adopt one of them, if any.

[paulmillar]

It might be interesting to compare the proposed approach to the I-ADOPT ontology. I think the equivalent I-ADOPT term would be Property. They have a List of Terminologies.