Re: draft response for 52b / JR6b

From: "Michael Schneider" <schneid@fzi.de>
Subject: RE: draft response for 52b / JR6b
Date: Tue, 24 Mar 2009 23:24:11 +0100

> Hi Peter!
> 
>>-----Original Message-----
>>From: Peter F. Patel-Schneider [mailto:pfps@research.bell-labs.com]
>>Sent: Monday, March 23, 2009 5:59 PM
>>To: Michael Schneider
>>Cc: ivan@w3.org; public-owl-wg@w3.org
>>Subject: Re: draft response for 52b / JR6b
>>
>>I have slightly lengthened this response to provide more background.
>>
>>See http://www.w3.org/2007/OWL/wiki/LC_Responses/JR6b
>>
>>peter
> 
> [[
> The direct semantics directly provides one meaning 
> for the constructed in OWL 2 ontologies.
> The RDF-based semantics directly provides a meaning for all RDF graphs. 
> As all OWL 2 ontologies can be mapped into RDF graphs, 
> the RDF-based semantics provides another semantics 
> for all the constructs in OWL 2 ontologies.
> ]]
> 
> I'm happy with this: "RDF-Based Semantics provides meaning for all RDF
> graphs."
> 
> The third sentence, while technically redundant, will certainly be
> helpful to people.
> 
> Btw, there is a typo in the first sentence: "constructed".

Fixed.

> [[
> The phrase "OWL 2", by itself, is now uniformly used 
> to refer to the entire language, regardless of syntax or semantics. 
> ]]
> 
> I'm not really sure what this means, in particular together with the 
> term "regardless of syntax". How can one talk about a language
> regardless of syntax? I probably miss something here.

How about "regardless of *the particular* syntax or semantics"?

> [[
> The phrase "OWL 2 Full", by itself, is now uniformly used as a shorthand to
> refer to the treatment of RDF graphs (particularly those RDF graphs that use
> OWL 2 constructs) under the RDF-based semantics and thus, as you say, is a
> combination of both syntax and semantics. This use of "OWL 2 Full" is
> consistent with the use of "OWL Full" in the WebOnt documents that define
> the original version of OWL.
> ]]
> 
> Fine: "OWL 2 Full" = Syntax and Semantics.
> 
> [[
> "OWL 2 DL ontologies" are then those OWL 2 ontologies that admit reasoning
> using well-known DL techniques when interpreted using the Direct Semantics,
> and that can be mapped to RDF graphs and back again without affecting their
> meaning in the direct semantics. This use of "OWL 2 DL ontologies" is
> consistent with the use of "OWL DL" in the WebOnt documents that define the
> original version of OWL." Section 3 of the OWL 2 Structural Specification
> provides a comprehensive and compact list of the extra conditions that are
> required for an OWL 2 ontology to be an OWL 2 DL ontology.
> ]]
> 
> I'm perfectly ok with talking about (the set of) "OWL 2 DL ontologies"
> whenever 
> talking about the "syntax things". However, I don't believe that the
> statement
> 
>   "OWL 2 DL ontologies is consistent with the use of 'OWL DL' in the 
>   WebOnt documents." 
> 
> is true (unless this was just a typo).

It is true in the sense that the DL part is the same.

> Here are two sentences that I found in S&AS Section 2, which claims to 
> be normative, and which is actually about the Abstract Syntax, so it's
> comparable with the Structural Spec document:
> 
> [[
> The abstract syntax is expressed both for this smaller language, 
> called the OWL Lite abstract syntax here, 
> and also for a fuller style of OWL, 
> called the OWL DL abstract syntax here.
> ]]

> [[
> OWL Lite and OWL DL closely correspond to the description logics 
> known as SHIF(D) and SHION(D)
> ]]
> 
> So there seemed to be a distinction between the "OWL XX abstract syntax"
> on the one side, and "OWL XX" as something that can correspond to
> some description logic. I am presuming here that SHOIN is not just 
> only a particular syntax.
> 
> Another hint to "OWL XX" = Syntax&Semantics, 
> preceding the first citation above:
> [[
> (Note, however, that both OWL DL and OWL Lite 
> do not provide all of the feature of RDF Schema.)
> ]]
> 
> I think this refers to the semantic aspects of RDFS and OWL XX, 
> that, for example, RDFS provides for metamodelling and OWL XX not.
> I don't see how this statement could be true if meant
> only for the syntactic aspects of RDFS and OWL XX.

"OWL DL" is used in various places in S&AS.  Perhaps the most
"authoritative" place that OWL DL occurs is in Section 5 - which is the
RDF-based semantics section!  However, I wouldn't want to say that in
OWL 1 "OWL DL" is generally thought of as pertaining to the RDF-based
semantics!

To be extraordinarily precise, one might say

  OWL 2 DL ontologies is consistent with the use of 'OWL DL ontologies'
  in the WebOnt documents."

or even

  OWL 2 DL is consistent with the use of 'OWL DL' in the WebOnt
  documents."

but I don't think that either change adds anything in particular to the
response.

> Nevertheless, I'm happy with the OWL 2 Full parts. In the meanwhile, the
> RDF-Based Semantics has changed to use the term "OWL 2 RDF-Based Semantics"
> everywhere where only the semantics is meant. There are two places where the
> old Full spec is cited, maybe this should also be changed (looks strange to
> compare the "OWL 2 RDF-Based Semantics" spec with the "OWL Full" spec).
> Also, I have replaced all the occurrences of "the OWL 2 Full universe" (was
> a Non-LC comment by Jonathan Rees). Now, at least much of the potential of
> confusion should have gone. There are still quite a lot terms such as "OWL 2
> Full interpretation", but they mirror the naming scheme of the old spec. 
> 
> Michael

peter

Received on Tuesday, 24 March 2009 23:31:16 UTC