Property without class in RDF

Question

I have a class Person. It has two properties hasAge and hasZipCode.

:Person :hasAge     <literal>
        :hasZipCode <literal>

I would like to say that hasAge and hasZipCode are sub-properties of hasColumn property. Hence I mentioned it like following.

:hasAge       rdfs:subPropertyOf :hasColumn
:hasZipCode   rdfs:subPropertyOf :hasColumn

However, in my RDF graph, I don't have any class that contains property hasColumn. I mean I don't have something like following:

:Table :hasColumn <literal>

Is it allowed in RDF to define a property (e.g., hasColumn) without a class?

Answer 1

I have a class ChicagoPerson. It has two properties hasAge and hasZipCode, and their values are mentioned below.

:ChicagoPerson :hasAge     "30"
               :hasZipCode "60543"

This doesn't make sense. In this snippet, ChicagoPerson is an individual (or resource), and has values for the two properties that you mention.

I would like to say that hasAge and hasZipCode are sub-properties of hasColumn property. Hence I mentioned it like following.

:hasAge       rdfs:subPropertyOf :hasColumn
:hasZipCode   rdfs:subPropertyOf :hasColumn

This is the correct way to do this. This means that if you have RDFS inference available, then you'll have the data:

:ChicagoPerson :hasAge     "30" ,
               :hasZipCode "60543" ,
               :hasColumn "30", "60543" .

However, in my RDF graph, I don't have any class that contains property hasColumn. I mean I don't have something like following:

:Table :hasColumn <literal>

Again, I think you're confusing the idea of classes and individuals. A class is a collection of individuals. If you have RDFS inference available, then the subproperty axioms you wrote above will allow that RDFS reasoner to infer the additional triples that I mentioned. The extra triples that you'll see are

:ChicagoPerson :hasColumn "30", "60543" .

not something with Table as the subject.

Is it allowed in RDF to define a property (e.g., hasColumn) without a class?

RDF doesn't really have any notion of defining properties. RDF just has triples of the form subject, predicate, object. Schema languages like RDFS and OWL assign special meaning to certain kinds of triples, usually by assigning a meaning to particular predicates. E.g., in RDFS and OWL, the predicate rdfs:subPropertyOf gets a special meaning.

Do you actually have an RDFS reasoner running on your dataset?