Why doesn't the Jena validator give me warnings? - jena

Question

In the following code, the ValidityReport is always valid; there is no inconsistency detected. I expect that it should give me a warning because I am giving a family name to a FOAF document. Why isn't an inconsistency detected?
Model model = ModelFactory.createDefaultModel();
OntModel m = ModelFactory.createOntologyModel();
m.read(FOAF.NS);
Resource persona = model.createResource("http://www.example.org/rdf#Persona", FOAF.Document);
persona.addProperty(FOAF.family_name, "18", XSDDatatype.XSDint);
InfModel infModel = ModelFactory.createRDFSModel(m, model);
ValidityReport validity = infModel.validate();
if (validity.isValid()) {
System.out.println("Valid!");
} else {
System.out.println("Conflicts");
for (Iterator<Report> in = validity.getReports(); in.hasNext();) {
System.out.println(" - " + in.next());
}
}

Answer 1

The instance data that you're creating is this:
#prefix ex: <http://www.example.org/rdf#> .
#prefix foaf: <http://xmlns.com/foaf/0.1/> .
#prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
ex:Persona a foaf:Document ;
foaf:family_name "18"^^xsd:int .
There's a resource of type foaf:Document, and it has an xsd:int value for the the foaf:family_name property.
Your inference model uses an RDFS reasoner. The domain of foaf:family_name (which, by the way, is described as an archaic spelling of foaf:familyName) is foaf:Person, so ex:Persona can thus be inferred to be a foaf:Person, as we see if we write the inference model. There are a number of other types that ex:Persona is inferred to have as well:
ex:Persona a foaf:Document , foaf:Person , <http://www.w3.org/2000/01/rdf-schema#Resource> , <http://www.w3.org/2003/01/geo/wgs84_pos#SpatialThing> , foaf:Agent ;
foaf:family_name "18"^^xsd:int .
It's rather hard to have inconsistencies in an RDFS model, since you can't really say a whole lot. You can't declare disjoint classes, for instance, so there's no contradiction between something being a foaf:Document and a foaf:Person.
Even if you use an OWL reasoner, you'd still need to pinpoint some particular logical inconsistency. I don't know whether any of those types that ex:Persona has are disjoint, and if they're not (or if the reasoner can't infer that they are), you're not going to find an inconsistency.