schema:domainIncludes vs rdfs:domain

[anatoly-scherbakov]

Hi,

I have a little question about the following segment of the vocabulary:

    {
      "@id": "skill:award",
      "@type": "rdf:Property",
      "rdfs:comment": "Award(s) given to the person (or a group of people) to recognize their merit or excellence.",
      "rdfs:label": "award",
      "schema:domainIncludes": {
        "@id": "skill:Resume"
      }
}

skill:award property can be attached to a skill:Resume instance, and is described with a schema:domainIncludes to describe that. Unfortunately, if the document author did not specify the skill:Resume type explicitly then we would be unable to reason that.

We could if we would define the domain like this:

"rdfs:domain": {
  "@id": "skill:Resume"
}

In that case, standard OWL rules will allow us to skip defining the type of a Resume object if we have an award defined for it.

Perhaps there are reasons as to why avoid this mechanism? I would be curious to learn about design motivations in this regard.

[ehsan6sha]

This is excellent. The idea is to get the great feedbacks from the developers' community and improve the standard. I'll also think about any reasons we didn't go this way. I need to dig into my memory and notes (wish I had put them here)

[anatoly-scherbakov]

Looking into schema.org vocabulary (which obviously defines schema:domainIncludes and schema:rangeIncludes, I can see numerous occasions similar to this:

schema:acceptedAnswer schema:rangeIncludes schema:Answer, schema:ItemList .

It describes situations more complicated than rdfs:range could but, for exchange, we no longer can employ inference.

I feel that we probably could convert schema:rangeIncludes to rdfs:domain in the cases where there is only one possible range specified for a given property.