SHACL shapes should rely on rdfs:subClassOf Reasoning

[VladimirAlexiev]

SHACL mandates rdfs:subClassOf Reasoning, and there are huge benefits if the shapes leverage it:
https://github.com/Sveino/Inst4CIM-KG/tree/develop/shacl-improved#leverage-rdfssubclassof-reasoning

Are there arguments not to rely on subClassOf Reasoning?
@Haigutus mentioned some people don't want to load the ontologies together with instance data: why?
The main purpose of the semantic web is so that data can be self-describing.

[griddigit-ci]

I think the only blocking point here is testing with different engines.
We did a version that relies on the subClassOf

There were 2 blocking points people were reporting

• performance - slow if on abstract, or
• just not supported

The objective is to go there one day as all will be much more efficient and we still have the possibility to put a constraint on a specific concrete class if that is required

[VladimirAlexiev]

Which engines were slow?
Should you twist ENTSOE constraints out of shape to accommodate some slow engines?
Shouldn't market competition take care of such performance problems?

If this is fixed, it will eliminate most checks for rdf:type eg
sh:path ( cim:AnalogLimit.LimitSet rdf:type )

CIM SHACL overuses such checks in a bad way:

grep "rdf:type )" */CurrentRelease/SHACL/*|wc -l
1085

Count per file:

grep -c "type )" */CurrentRelease/SHACL/*|grep -v :0
CGMES/CurrentRelease/SHACL/61970-456_StateVariables-AP-Con-Complex-Explicit-CrossProfile-SHACL.ttl:1
CGMES/CurrentRelease/SHACL/61970-552-Header-AP-Con-Simple-SHACL.ttl:5
CGMES/CurrentRelease/SHACL/61970-600-2_DiagramLayout-AP-Con-Simple-SHACL.ttl:7
CGMES/CurrentRelease/SHACL/61970-600-2_Dynamics-AP-Con-Simple-SHACL.ttl:87
CGMES/CurrentRelease/SHACL/61970-600-2_Equipment-AP-Con-Simple-SHACL.ttl:77
CGMES/CurrentRelease/SHACL/61970-600-2_EquipmentBoundary-AP-Con-Simple-SHACL.ttl:11
CGMES/CurrentRelease/SHACL/61970-600-2_GeographicalLocation-AP-Con-Simple-SHACL.ttl:8
CGMES/CurrentRelease/SHACL/61970-600-2_Operation-AP-Con-Simple-SHACL.ttl:18
CGMES/CurrentRelease/SHACL/61970-600-2_StateVariables-AP-Con-Simple-SHACL.ttl:1
CGMES/CurrentRelease/SHACL/61970-600-2_Topology-AP-Con-Simple-SHACL.ttl:4
NCP/CurrentRelease/SHACL/AssessedElement-AP-Con-Simple-SHACL.ttl:15
NCP/CurrentRelease/SHACL/AvailabilitySchedule-AP-Con-Simple-SHACL.ttl:13
NCP/CurrentRelease/SHACL/Contingency-AP-Con-Simple-SHACL.ttl:7
NCP/CurrentRelease/SHACL/DatasetMetadata-AP-Con-SHACL.ttl:13
NCP/CurrentRelease/SHACL/EquipmentReliability-AP-Con-Simple-SHACL.ttl:261
NCP/CurrentRelease/SHACL/GridDisturbance-AP-Con-Simple-SHACL.ttl:98
NCP/CurrentRelease/SHACL/ImpactAssessmentMatrix-AP-Con-Simple-SHACL.ttl:15
NCP/CurrentRelease/SHACL/MonitoringArea-AP-Con-Simple-SHACL.ttl:7
NCP/CurrentRelease/SHACL/ObjectRegistry-AP-Con-Simple-SHACL.ttl:6
NCP/CurrentRelease/SHACL/PowerSchedule-AP-Con-Simple-SHACL.ttl:10
NCP/CurrentRelease/SHACL/PowerSystemProject-AP-Con-Simple-SHACL.ttl:3
NCP/CurrentRelease/SHACL/RemedialAction-AP-Con-Simple-SHACL.ttl:71
NCP/CurrentRelease/SHACL/RemedialActionSchedule-AP-Con-Simple-SHACL.ttl:23
NCP/CurrentRelease/SHACL/SecurityAnalysisResult-AP-Con-Simple-SHACL.ttl:10
NCP/CurrentRelease/SHACL/SensitivityMatrix-AP-Con-Simple-SHACL.ttl:7
NCP/CurrentRelease/SHACL/StateInstructionSchedule-AP-Con-Simple-SHACL.ttl:57
NCP/CurrentRelease/SHACL/SteadyStateHypothesisSchedule-AP-Con-Simple-SHACL.ttl:77
NCP/CurrentRelease/SHACL/SteadyStateInstruction-AP-Con-Simple-SHACL.ttl:175

Once eliminated, we can focus on tricky/exoting things like Sveino/Inst4CIM-KG#15

[VladimirAlexiev]

It will also eliminate most uses of sh:in:
https://github.com/Sveino/Inst4CIM-KG/tree/develop/shacl-improved#dont-overuse-shin.
Most of these are disjunctive type checks:

 grep "sh:in " */CurrentRelease/SHACL/*|grep -v '"'|wc -l
656

Only these are literal checks (eg CGMES profile anyURIs, days of the week, etc)

grep 'sh:in .*"' */CurrentRelease/SHACL/*|wc -l
18