proposal to change representation of attributes and units

VladimirAlexiev commented 4 months ago

Currently datatypes are defined like this:

cim:ActivePower  rdf:type       rdfs:Class ;
        rdfs:comment            "Product of RMS value of the voltage and the RMS value of the in-phase component of the current." ;
        rdfs:label              "ActivePower"@en ;
        cims:belongsToCategory  er:Package_EquipmentReliabilityProfile ;
        cims:stereotype         "CIMDatatype" .

cim:ActivePower.multiplier
        rdf:type           rdf:Property ;
        rdfs:domain        cim:ActivePower ;
        rdfs:label         "multiplier"@en ;
        rdfs:range         cim:UnitMultiplier ;
        cims:isFixed       "M" ;
        cims:multiplicity  cims:M:0..1 ;
        cims:stereotype    <http://iec.ch/TC57/NonStandard/UML#attribute> .

cim:ActivePower.unit  rdf:type  rdf:Property ;
        rdfs:domain        cim:ActivePower ;
        rdfs:label         "unit"@en ;
        rdfs:range         cim:UnitSymbol ;
        cims:isFixed       "W" ;
        cims:multiplicity  cims:M:0..1 ;
        cims:stereotype    <http://iec.ch/TC57/NonStandard/UML#attribute> .

cim:ActivePower.value
        rdf:type           rdf:Property ;
        rdfs:domain        cim:ActivePower ;
        rdfs:label         "value"@en ;
        cims:dataType      cim:Float ;
        cims:multiplicity  cims:M:0..1 ;
        cims:stereotype    <http://iec.ch/TC57/NonStandard/UML#attribute> .

And then attributes (data props) are defined like this:

cim:GeneratingUnit.maxEconomicP
        rdf:type           rdf:Property ;
        rdfs:domain        cim:GeneratingUnit ;
        cims:dataType      cim:ActivePower ;
        cims:stereotype    <http://iec.ch/TC57/NonStandard/UML#attribute> .

There are several problems:

the data prop doesn't define a range (while the value of GeneratingUnit.maxEconomicP in instance data is a a pure number)
cim:ActivePower.unit and cim:ActivePower.multiplier are never used
cim:Float is a non-standard datatype (eg xsd:double is a standard datatype)

Proposals:

[x] Define range of attributes:

cim:GeneratingUnit.maxEconomicP rdfs:range xsd:double.

[x] Define them as datatype props. Using this OWL class doesn't mean you're going away from RDFS, you're just being more specific about the nature of the property
```
cim:GeneratingUnit.maxEconomicP a owl:DatatypeProperty.
```
[x] Remove cim:ActivePower.unit and cim:ActivePower.multiplier since they are not used (but if you like them, you can keep them)
[x] Attach the unit and multiplier to cim:ActivePower, eg
```
cim:ActivePower 
cim:multiplier "M";
cim:unit "W".
```

[x] Attach the unit and multiplier to the attribute, eg

cim:GeneratingUnit.maxEconomicP 
cim:multiplier "M";
cim:unit "W".

[x] Switch to QUDT. The string values above are not very good since they can't carry descriptive info. So you can replace the above with
```
cim:ActivePower qudt:unit unit:MW.
cim:GeneratingUnit.maxEconomicP qudt:unit unit:MW.
```

VladimirAlexiev commented 4 months ago

EquipmentReliability-AP-Voc-RDFS2020.ttl has a list of units, eg:

cim:UnitSymbol.W  rdf:type  cim:UnitSymbol ;
        rdfs:comment     "Real power in watts (J/s). Electrical power may have real and reactive components. The real portion of electrical power (I&#178;R or VIcos(phi)), is expressed in Watts. See also apparent power and reactive power." ;
        rdfs:label       "W"@en ;

cim:UnitMultiplier.M  rdf:type  cim:UnitMultiplier ;
        rdfs:comment     "Mega 10**6." ;
        rdfs:label       "M"@en ;
        cims:stereotype  "enum" .

and then a shape in EquipmentReliability-AP-Con-Simple-SHACL.ttl asks a certain property to use these enumerated instances:

er:Curve.y1Unit-datatype
        rdf:type        sh:PropertyShape;
        sh:path         cim:Curve.y1Unit;
        sh:in           ( cim:UnitSymbol.A cim:UnitSymbol.none cim:UnitSymbol.VPerVAr cim:UnitSymbol.WPerm2 cim:UnitSymbol.s cim:UnitSymbol.Pa cim:UnitSymbol.WPers cim:UnitSymbol.deg cim:UnitSymbol.degC cim:UnitSymbol.VAr cim:UnitSymbol.Hz cim:UnitSymbol.V cim:UnitSymbol.W cim:UnitSymbol.ohm );

So instead of "M" and "W" above, everywhere we should use cim:UnitMultiplier.M and cim:UnitSymbol.W respectively. That includes cim:ActivePower.multiplier, cim:ActivePower.unit! These enumerated instances still carry less descriptive info than QUDT unit:* but we can stick with them.

VladimirAlexiev commented 3 months ago

This is very similar to https://github.com/3lbits/CIM4NoUtility/issues/338 but addresses some other aspects.

VladimirAlexiev commented 2 months ago

https://github.com/Sveino/Inst4CIM-KG/issues/38 describe this in more detail, and rdfs-improvement/README in even more detail:

Datatypes and Units of Measure

The last sec has mapping tables to QUDT.

So let's NOT discuss here, let's discuss there. But this issue has a task list, so I'll close it only after everything is done.

Sveino / Inst4CIM-KG

proposal to change representation of attributes and units #29