`xsd:anyURI` vs URI

[VladimirAlexiev]

The SHACL shape and https://ns.mh1.eu/epcis/ show xsd:anyURI as range for several properties. However, xsd:anyURI is a literal not a URI.

Instead, these props:

• should be defined as owl:ObjectProperty in the ontology
• should have a defined range class, eg
  • gs1:Place (for SGLN),
  • a superclass gs1:PhysicalObject (for EPC) that includes gs1:Product, gs1:LogisticalUnit, etc etc
• don't need to be described with sh:nodeKind sh:BlankNodeOrIRI in the shape, because owl:ObjectProperty can only have such values

[VladimirAlexiev]

This script makes the prop list below.

riot -out nt EPCIS.ttl | perl -ne 'print if s{.*epcis/(.*?)>.*rangeIncludes.*anyURI.*}{1. $1}'

I've added suggested ranges, conforming to EPCIS fields to classes; added a few at the bottom

1. bizRules: ?
2. bizStep: cbv:BizStep
3. bizTransaction: epcis:BizTransaction or epcis:BizTransactionDocument (#215)
4. chemicalSubstance: ?
5. childEPCs: gs1:PhysicalObject
6. correctiveEventIDs: epcis:EPCISEvent
7. dataProcessingMethod: ?
8. destination: epcis:SourceOrDestination
9. deviceID: gs1:SensorDevice ?
10. deviceMetadata: ?
11. disposition: cbv:Disp
12. epcList: gs1:PhysicalObject
13. id: none, pseudo-prop
14. inputEPCList: gs1:PhysicalObject
15. microorganism: ?
16. outputEPCList: gs1:PhysicalObject
17. parentID: gs1:PhysicalObject, gs1:Place ?
18. rawData: ?
19. set: cbv:Disp
20. source: epcis:SourceOrDestination
21. unset: cbv:Disp
22. uriValue: ?. additions are below this line
23. epcClass: gs1:Product
24. quantity: xsd:double
25. eventList: epcis:EPCISEvent
26. epcis:correctiveEventIDs: epcis:EPCISEvent
27. epcis:errorDeclaration epcis:ErrorDeclaration
28. epcis:reason: cbv:ER
29. epcis:measurementType: gs1:MeasurementType
30. epcis:bizTransactionType: cbv:BTT,
31. epcis:sourceOrDestinationType: cbv:SDT

For "?" we have a choice:

• a: DatatypeProperty with type anyURI (i.e. a string with a datatype)
• b: ObjectProperty without a class (i.e. a "free" URL). That's my preference
• c: ObjectProperty with a class (eg gs1:Microorganism). But that would be a mere marker class (with no props defined in GS1 itself), so not very useful

The CBV classes should be considered "extensible", i.e. a user is able to add a new class member (they are anyway, but we should mention it in the section "EPCIS semantics")

@mgh128 please comment by item

[VladimirAlexiev]

@mgh128 @CraigRe @RalphTro Can we make a decision on this one? We're ready to start implementing it (@nikolatulechki, or @nicksinch)

[CraigRe]

@VladimirAlexiev what are you ready to start implementing? None of this has been released to public community review yet, let alone published as a standard...

[VladimirAlexiev]

@CraigRe Implement changes to the ontology.

[VladimirAlexiev]

ontology-prop-checks.rq makes about 10 checks on props:

• mismatching rdfs:domain and schema:domainIncludes
• mismatching rdfs:range and schema:rangeIncludes
• rdf:Property but not owl:ObjectProperty or owl:DatatypeProperty and vice versa
• owl:DatatypeProperty but range is not a xsd: datatype
• owl:DatatypeProperty with range xsd:anyURI should be changed to owl:ObjectProperty, see #206

Only the last one has problems, they are down from 31 to 20:

epcis:bizRules epcis:bizStep epcis:bizTransaction epcis:chemicalSubstance epcis:childEPCs epcis:correctiveEventIDs epcis:dataProcessingMethod epcis:deviceID epcis:deviceMetadata epcis:disposition epcis:epcList epcis:id epcis:inputEPCList epcis:microorganism epcis:outputEPCList epcis:parentID epcis:rawData epcis:set epcis:unset epcis:uriValue

[VladimirAlexiev]

21fe79d (#270) fixes this. @mgh128 please:

• merge the PR
• assign to yourself to fix the web app appropriately (eg uriValue is ObjectProperty with no range)