improve class and property definitions

[VladimirAlexiev]

@philarcher Successful ontologies pay attention to proper definitions.

• See https://github.com/gs1/EPCIS/issues/207 for guidance. @justin2004, related to #25, what do you think of that guidance?
• See https://github.com/gs1/EPCIS/issues/208, https://github.com/gs1/EPCIS/issues/211 for examples of problematic definitions in EPCIS

Examples from WebVoc:

• A Country is not a data structure https://github.com/gs1/WebVoc/issues/25 , nor is it restricted to commercial activities
• A Website URL is not a Simple text string https://github.com/gs1/WebVoc/issues/26
• https://www.gs1.org/voc/additionalOrganizationIdentificationTypeValue "The value associated with the Additional Organization Identification Type Value": let me wrap my head around this: "The value associated with the value"... Nope, doesn't still doesn't make sense

[VladimirAlexiev]

Conceptual problems like this in #25 can be avoided, if you describe things for what they are, not for what they appear to be when viewed from piecemeal data integration efforts burdened by backward compatibility considerations. See https://github.com/gs1/EPCIS/issues/207 for guidance ("Ontological Realism").

Another example:

https://www.gs1.org/voc/Organization, in general, looks good. But what is the field gs1:department doing in there? Are we describing an Org or a Department?

• Can I use multiple values of gs1:department (then what do the contact data refer to?)
• or must I use several instances of gs1:Organization (then those nodes don't describe Orgs but OrganizationalUnits)?
• How do I make a link from Org to Department?

The W3 Org ontology describes these things clearly (Organization, OrganizationalUnit, hasUnit, unitOf). This model was reached after much discussion at EU SEMIC, W3C, etc. @philarcher was strongly involved afaik. Why didn't WebVoc adopt these patterns, but made its own, with its own mistakes?

[justin2004]

@VladimirAlexiev I think your quotes from that guidance are right on. Though I still need to look at the source papers more.

Just a side note about this:

The reason is that the EPCIS XML (and JSON) representations are data centric, which influences the design of RDF classes and props. Please see Ontological Realism as a Methodology for Coordinated Evolution of Scientific Ontologies, which says

    an ontology should be analogous not to a data model, but rather to a reality model

that is one sense of "data centric" but there is also this sense of "data centric" and I see the latter sense in use quite a bit.

[VladimirAlexiev]

@justin2004 I did sign this manifesto a year ago http://datacentricmanifesto.org/sign/1497/. Basically it says "your data is more important than the apps and systems that process it".

What I add is "your data, harmonized and semantically expressed, is much more valuable than stovepiped data developed in isolated efforts". That's why I quipped at signing "Semantic Data Integration will set you free!"