Re: issue-118 proposed text

On Fri, May 03, 2013 at 11:48:12AM -0500, Pat Hayes wrote:
> > On Wed, May 01, 2013 at 10:23:06AM -0500, Pat Hayes wrote:
> >> If D is a set of IRIs which are used to refer to datatypes, then the elements
> >> of D are called <em>recognized</em> datatype IRIs.
> > 
> > Let's say I have a set of IRIs which are used to refer to datatypes:
> > 
> >    http://purl.org/dc/terms/ISO3166
> >    http://purl.org/dc/terms/ISO639-2
> >    http://purl.org/dc/terms/ISO639-3
> > 
> > I have decided to call the set "D". 
> > 
> > Does this make them "recognized" datatype IRIs?
> 
> They are recognized by you if you think they are. Presumably, you also think
> they denote datatypes, and hopefully you have made public what datatypes you
> intend them to denote. Even more hopefully, you have made it possible to find
> that out by following them as links, so that people reading your RDF can also
> treat them as recognized and draw useful conclusions which involve your
> datatype. But all this hopeful stuff is outside the RDF Semantics
> specification, and so all we can do there is hope that that you enable your
> readers to share in your recognition. If they don't, then the semantics they
> apply will not recognize those IRIs and they will not be able to draw as many
> conclusions as you can. 
> 
> Put another way, the effective semantics of some RDF might vary depending on
> how many IRIs you are able to recognize. That is why it treats D as a
> parameter.
> 
> Does that make sense? In the 2004 semantics this was all described in terms
> of datatype maps, but the same issues applied, since if I read your RDF and
> didn't know your datatype map then I couldn't apply it in my reasoning. 

Hi Pat,

Yes, thank you - this does make sense, and I had correctly understood this to
be the case.  

I raised this because the text quoted above is meant to be added to Section 5.4
of Concepts.  However, without the context provided by Semantics, the point
seems a bit... mysterious.  What is a reader of Concepts who has not already
read Semantics (i.e., most readers of Concepts), to make of the notion that
datatype IRIs can be "recognized"?  This could be fixed by adding a bit more
explanation to Concepts, perhaps with a "see also" pointing to Semantics.

Tom

-- 
Tom Baker <tom@tombaker.org>

Received on Saturday, 4 May 2013 20:50:28 UTC