RE: ISSUE-142 (Rdlrelation): Problems with statement re: relationship between OWL-RL DL and OWL-RL Full

Hello,

I have a few questions about this issue inline.

> -----Original Message-----
> From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org] On Behalf Of OWL Working
> Group Issue Tracker
> Sent: 15 August 2008 21:43
> To: public-owl-wg@w3.org
> Subject: ISSUE-142 (Rdlrelation): Problems with statement re: relationship between OWL-RL DL and OWL-
> RL Full
> 
> 
> 
> ISSUE-142 (Rdlrelation): Problems with statement re: relationship between OWL-RL DL and OWL-RL Full
> 
> http://www.w3.org/2007/OWL/tracker/issues/142
> 
> Raised by: Alan Ruttenberg
> On product:
> 
> http://www.w3.org/2007/OWL/wiki/Profiles#Relationship_between_OWL-R_DL_and_OWL-R_Full
> 
> Spec says:
> "let O be an OWL-R DL ontology in which no URI is used both as an object and a data property; and let
> F be a set of assertions of the following form"
> 
> But O already includes assertions like those in F and more. What's meant is O is an ontology in
> something less than OWL-R DL.
> 

I really don't understand this comment.

I hope, however, that an explanation of this issue might help you. Here is how you should read it:

- O can be any OWL-R DL ontology.
- The ontology O should not use punning. (That's the meaning of the statement "no URI is used...".)
- F should be an ontology that contains assertions only.

In this case, O entails F iff and only if the RDF encoding of O entails the RDF encoding of F under the rule semantics.

I hope this makes things clearer. To further clarify things, note that F should not contain axioms such as SubClassOf. In this case,
it is not guaranteed that the equivalence holds. To make things clearer: in some cases the equivalence will hold, but in some cases
it won't. I wasn't able to formally characterize the good cases; therefore, I decided to take the safe option and simply state the
correspondence for that I know holds.

> Also, it is unclear what happens in the case of datatype inconsistencies. (I think nothing in the
> case of the rules). So it is unclear to me how to interpret the condition
> 
> F is a consequence of O under the OWL 2 DL semantics if and only if RDF(F) is a consequence of RDF(O)
> ? AXIOMS under the standard first-order semantics.
> 
> In cases such as: PropertyDomain(p1 xsd:string) PropertyAssertion(p1 x "1"^^xsd:int)), as in OWL-DL
> everything follows from such an inconsistency, but this isn't inconsistent according with the rules
> currently stated.
> 

The current set of rules doesn't say anything the semantics of datatypes, and that's a well-known TODO item for OWL RL: the Profiles
document already contains Jeremy's comment that something needs to be said about datatype reasoning. 

I haven't addressed this comment because at first it wasn't clear to me how to address the issue. This is not completely trivial if
you want to preserve the monotonic nature of the logic. I've been thinking about the problem, however, and I think I know now how to
address it. I haven't had the time to do so yet; however, I wanted to address all reviewers' comments in the next month of so. In
particular, I was waiting for a resolution to the unification issue: this will require the OWL RL part of the Profiles document to
change significantly, and this would then give me the opportunity to address all reviewers' comments, including the datatypes issue.

Regards,

	Boris

> 
> 
>

Received on Saturday, 16 August 2008 08:12:02 UTC