Re: Status of OWL 2 Mapping to RDF Graphs

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Subject: Re: Status of OWL 2 Mapping to RDF Graphs
Date: Mon, 18 May 2009 16:16:59 -0500

> On Mon, May 18, 2009 at 4:55 PM, Peter F.Patel-Schneider
> <pfps@research.bell-labs.com> wrote:
>> From: Alan Ruttenberg <alanruttenberg@gmail.com>
>> Subject: Re: Status of OWL 2 Mapping to RDF Graphs
>> Date: Mon, 18 May 2009 15:28:50 -0500
>>
>>> On Mon, May 18, 2009 at 4:16 PM, Peter F.Patel-Schneider
>>> <pfps@research.bell-labs.com> wrote:
>>>> I'm not sure what part of the documents you are basing this behaviour
>>>> on.   As far as I can see our documents leave it completely up to
>>>> implementations as to how they are to implement the relationship between
>>>> rdf:text literals and plain literals, and this is how it should be.
>>>
>>> In this message from Sandro he asserts that it was a requirement of HP
>>> for moving to last call that rdf:text literals do not "escape" into
>>> RDF.
>>>
>>> http://lists.w3.org/Archives/Public/public-rdf-text/2009AprJun/0077.html
>>>
>>> -Alan
>>
>> Sure, but why should this affect any document except the rdf:text
>> document?  Our relevant documents defer to rdf:text, which is only
>> appropriate.
>>
>> peter
>>
> 
> My reasoning is that the RDF mapping is a very precise document and
> doesn't address this aspect of the translation, whereas it seems to be
> in scope. 

I don't think that this is in scope of the RDF mapping document at
all.  In fact, I think that this is specifically out of scope of that
document!  

> It looked to me like there could be confusion that on the
> one hand the RDF mapping seems to say literals are untouched in
> translation and on the other hand there is advise [sic] in a citation that
> indicates otherwise. 

I disagree.  The Mapping to RDF document indicates how OWL tools should
translate between the Functional Syntax and RDF Graphs (which is to
leave literals alone, as is appropriate).  The rdf:text document says
what should be done when moving RDF Graphs around (which is to replace
rdf:text literals with plain literals).  These are two different
things.  

Of course a tool may decide to translate between the Functional Syntax
and an RDF Graph for the purposes of exchange, in which case it may as
well do the rdf:text replacement when it does the translation to the RDF
graph, but this is an optimization, and not something that our documents
should spend time worrying about.

> Someone who will implement a translation will, I
> expect, be looking at the RDF mapping document closely and miss this,
> having consequence for users who are not expecting their string
> literals to [not?] change in this way.

Well, then you were not doing your homework.  If you use rdf:text then
you are supposed to be looking at the rdf:text document, as is
specifically mentioned in our Syntax document (in Section 4).

Is the mapping document supposed to say something about RDF graphs that
are not serializable in RDF/XML?  Is the mapping document supposed to
say something about the details of Turtle, or n3, or any other RDF
serialization?  I don't think so, and I don't think that it should be
saying anything about the requirements for exchange of rdf:text
literals.

> -Alan

peter

Received on Monday, 18 May 2009 23:15:05 UTC