Search Mailing List Archives
[protege-owl] Represent facts about facts???
harweis at gmail.com
Tue Feb 17 10:19:37 PST 2009
On Tue, 2009-02-17 at 17:56 +0000, Matthew Horridge wrote:
> In the latest OWL 2 specification (not yet supported by Protege 4, but
> we're working on it) the base triple is included in the RDF
> serialisation. The triple is also "reified" using owl:subject,
> owl:predicate and owl:object triples.
Thank you for that information - I assume then, that this duplication
should indeed be without any issues?
> Anyway, if you query your ontology using Pellet as the query engine,
> then I would have thought that you wouldn't need to worry about the
> triples in the RDF graph representation of the asserted information.
The issue for now is that we want to at least for now remain to be
compatible and usable in a simple Sesame RDF store.
In any case the approach of combining the base triple on the one hand
and the reified version for representing the annotation on the other
hand should solve most of what we want to achieve.
One problem I found by experimenting a bit with Protege4 (I am using
build 108) is this:
* I created a tiny ontology with just two classes, one object property,
two individuals and one axiom :Ind1 :hasProp1 :Ind2
* I added an annotation for the that whole axiom
* I stored the ontology as RDF/XML and re-loaded it: the axiom was
gone and the annotation now appeared for :Ind2 types :Class2 !
* The same happened when I stored and reloaded in Turtle syntax.
The turtle syntax generated for this is - to my knowledge and according
to the Sesame2 parser - not even in correct syntax:
> :Ind1 rdf:type :Class1 .
> [ rdf:type owl:Axiom ;
> rdf:subject :Ind1 ;
> rdf:predicate :hasProp1 ;
> rdf:object :Ind2 ;
> :hasSource "The Source1"
> ] .
> :Ind2 rdf:type :Class2 .
The [ ] stands by itself which I think is not allowed as [ ] can only
appear in the place of a subject or object, no?
More information about the protege-owl