`epcis:Object` class hierarchy

[VladimirAlexiev]

• The epcis:EPCISEvent class hierarchy is well-rounded 1) IMHO we need a class epcis:Object, to be the range of epcList (which is not xsd:AnyURI, see #206).
  • Although an event doesn't carry the types of related objects, the form of TDS identifier dictates it 2) Should it be called the above, or epcis:EPCISObject? The capitalized EPCIS is redundant but is already present in epcis:EPCISEvent. For consistency, we should either remove it from Event, or add it to Object. 3) We should also make a hierarchy of all subclasses corresponding to TDS identifiers.

We need to decide a class for each TDS identifier, see https://github.com/VladimirAlexiev/EPCIS/blob/ontology/Ontology/identifier-object-mapping.md:

• maps from TDS identifiers to classes
• lists the sources I used
• proposes a hierarchy

Issues: 4) How to distinguish gs1:locationGLN (AI 414) vs gs1:partyGLN (AI 417) as stated in Digital Link Semantics?

• Is this distinction the same as urn:epc:id:sgln: vs urn:epc:id:pgln: as per TDS? 5) How to reconcile epcis:BizLocation (used for any GLN) with gs1:Place, gs1:Organization (used for location vs party GLN)? 6) Where is epcis:ReadPoint in this hierarchy? Elsewhere I argued that epcis:ReadPoint may not be needed 7) BIC, ADI, DOC, IMOVN are missing in Digital Link Semantics 8) GMN, SRIN are missing in TDS 1.13, are they defined in a subsequent version? 9) Can you add comments where missing? I have an incomplete understanding of a good number of these identifiers 10) GTIN LGTIN SGTIN UPUI all map to gs:Product, but with additional attribute requirements (Eg UPUI must have gs1:hasThirdPartyControlledSerialNumber). Is this ok? It contradicts the "individual vs class" distinction 11) Should we use gs1: classes as specified in "Digital Link Semantics", or make our own 12) Master Data "vClasses" should be eliminated. Eg when you describe a BizLocation, you should use the same class as when referring to it from an event. I.e. the incoming link and outgoing links/attributes of a BizLocation should relate to the same class. 13) Should the root epcis:Object be in one namespace, whereas all subclasses are in another (gs1:)? 14) Can't be a flat hierarchy, eg need IndividualProduct (physical object), vs Product (product class_, vs Place vs Organization

[mgh128]

If we work on this, I think we'd name it epcis:EPC (for instance identifiers) and epcis:EPCClass (for any class-level identifiers, e.g. GTIN, GTIN+Lot etc.)

AI (414) corresponds to urn:epc:id:sgln and gs1:locationGLN AI (417) corresponds to urn:epc:id:pgln and gs1:partyGLN

BIC, ADI, DOC, IMOVN are missing in GS1 Digital Link Semantics because they are not Class 1 or Class 2 GS1 identification keys.

TDS v1.13 is the current version. There are plans to update TDS but not primarily motivated by the absence of GMN and SRIN.

I think this discussion is mostly for the GS1 Digital Link MSWG and we need to address some gaps or perhaps think of enhancements for that.

Most of the EPCIS 2.0 MSWG have not expressed deep interest in RDF triples so far. They support GS1 Digital Link URI as a Web-friendly identifier that can link to online information, services, master data etc. in a far more straightforward way than starting with an EPC URN. It seems to me that unfortunately very few (you, me, José) are deeply concerned with RDF semantics.

[VladimirAlexiev]

EPC ("code") is an identifier, not the thing being identified (product, product class, container, etc). So the class name should be Object or Entity.

BizLocation should not be in the same hierarchy and should have subclasses gs1:Place (sgln) and gs1:Organization (pgln): since these are not Objects.

And now you see the difficulty with ReadPoint: is it a gs1:Place or an gs1:Organization? Neither, because either of these classes can play the role of readPoint.

Furthermore: the vXxxxx classes used in Master Data must be unified with (merged to) these classes. JSON doesn't need to change, we'll map them using the context.

[VladimirAlexiev]

• numbered the issues on top
• see https://github.com/VladimirAlexiev/EPCIS/blob/ontology/Ontology/identifier-object-mapping.md draft

[VladimirAlexiev]

6: There won't be a superclass BizLocation nor ReadPoint, instead the ontology should use owl:unionOf (gs1:Place gs1:Organisation)

[VladimirAlexiev]

@CraigRe @mgh128 @RalphTro Most decisions (incl class names) are made in class hier.

• A few remain (eg 7,8) in EPC-class mapping, see crossed-out
• These involve no effort, but who'll make the final call about them?
• If you make a change, let us know to edit the ontology.

12 is a major issue that I'm not sure we've tackled. Craig, maybe you should make a separate issue about it?