Re: OWL 1.1 and the RDF heritage

[CC'd to public-owl-dev, where OWL1.1 is discussed in the first place]

On 27.02.2007 20:46, Michael Erdmann wrote:

> List,
> can someone please explain the following sentence in the current OWL 1.1 
> Semantics draft [1]. The abstract reads:
>> [...] OWL 1.1 does not provide an RDF-compatible semantics; ontologies 
>> encoded in OWL RDF should be interpreted by converting them into 
>> functional-style syntax and interpreting the result as specified in 
>> this document.
> Does this mean, there will be no RDF/XML serialization for OWL1.1 
> models, or will OWL 1.1 merely drop the pseudo-compatibility between OWL 
> and RDF Schema?
> thanks,
>   michael
> [1]
> -- 
>  email:		Dr. Michael Erdmann
>    tel: +49 / 163 / 665 7929
>  Registered Office:  Karlsruhe, Germany, HRB 9540
>  Managing Directors: Prof. Dr. Juergen Angele, Hans-Peter Schnurr

Hi, Michael!

This is how I currently understand this (pretty new) situation. I hope, 
others here in the list will correct me wherever I'm wrong.

OWL 1.1 introduces "semanticless" comments, which can be used to 
annotate all kinds of axioms. Here is an example from [1, 3], given in 
abstract syntax:

   SubClassOf( Comment("Humans are a type of animals.") Human Animal)

which is meant to be semantically equivalent to

   SubClassOf( Human Animal )

I see two problems with such kinds of comments:

First, an RDF mapping problem. It is not immediately clear to me how to 
map a construct like the above commented subclass axiom to RDF. It 
should probably be something like (in N3):

   <URIofAxiom> owl11:comment "a literal" .

But what is the URI of an /axiom/? Seen from an RDF perspective, an OWL 
axiom is a set of RDF statements, aka an RDF graph. And at least 
currently, there is no official spec for handling RDF graphs as RDF 
resources, which can have an URI. (There is, however a proposal on this, 
see [2]).

Second, a semantical problem. Say, we restrict our usage of such 
comments to only annotate /resources/. The problem is, that an RDF 
triple like

   <URIofResource> owl11:comment "a literal" .

cannot be really semantic-less. It is always assumed in RDF(S) that 
there is some interpretation for a given RDF triple, derived from the 
interpretations of its subject, predicate and object. According to [3, 

   "The basic intuition of model-theoretic semantics is that asserting
   a sentence makes a claim about the world: it is another way of
   saying that the world is, in fact, so arranged as to be an
   interpretation which makes the sentence true."

This would not mean, of course, that one cannot define a /syntactical/ 
mapping to RDF, but such a mapping would then not preserve semantics. 
And, together with my first point, there really seems to be not even a 
complete syntactical RDF-mapping for OWL1.1, but I might be wrong in this.


[1] OWL1.1 Structural Specification and Functional-Style Syntax

[2] Carrol, Bizer, Hayes, Stickler: "Named Graphs"

[3] RDF Semantics

Received on Tuesday, 27 February 2007 23:06:28 UTC