Re: Is N-Triples Normative? [Was: Re: model theory (W3C publication stuff)]

>I understand the points you guys are making.  However, to add a related
>issue, the current RDF/XML Syntax draft says "This process [that is,
>mapping from the XML syntax to the RDF model] is not yet complete, in
>that the final step is defining for each syntax production which RDF
>statements are added to the resulting model (if any)."  How are we going
>to describe "which RDF statements are added to the resulting model" for
>these productions?  Pictures of graph snippets?  If we use Ntriples
>here, and other places where it appears we want to be *really* precise
>(like in test cases), it sure seems funny to say it isn't normative.

I disagree. By refusing to acknowledge 'normativity' we are just 
declaring a refusal to get involved with a whole lot of 
'standardising' effort. If anyone doesnt like Ntriples they can draw 
the graphs themselves. Ntriples isnt really to do with RDF as such: 
its a graph-linearisation convention and would work for any labelled 
graph.

> On
>the other hand, what we *really* mean to be normative here is the
>translation from XML/RDF to RDF, not any particular way of representing
>the RDF.  (Would saying we describe RDF statements in terms of
>particular collections of binary predicates in mathematical logic be any
>better?  That's more or less what Ntriples are anyway, and may sound
>less like another "syntax").

I would prefer not to do that, since that would be taken by many folk 
as importing a 'standard' semantics (which is in fact slightly 
different from the one we have right now...)

> In any event, given the current state of
>affairs, we need to be damn careful that we understand all the ins and
>outs of the various graphs and their mappings to Ntriples.

I agree. We can just publish Ntriples and describe it honestly as a 
NON-normative, but extremely handy, notation for describing RDF 
graphs in linear text, and *that's all*. Then we can use it in our 
prose freely, and the world will be able to follow us pretty easily 
once they get used to it (particularly if the document gives some 
examples and is careful to say that this is not being offered as an 
alternative RDF serialization - though of course everyone will see 
that it could be used that way, but we aren't *saying* that) and yet 
we won't be asked to give a semantics for it, write parsers for it, 
provide an XML serialization of it, etc. etc. . Put the BNF in an 
appendix in case anyone wants to use it, but don't wave any banners. 
(To rub the point in, we could even characterise it as a language for 
describing *pictures* of RDF graphs, which is how I tend to think of 
it in any case.)

I think the following is true, BTW: disregarding ordering of lines in 
an Ntripledoc and renamings of <anonNode>s , Ntripledocs and rdf 
graphs are 1:1;  every doc describes a unique graph, and every graph 
has a unique doc, and if you start with one, translate into the 
other, then back, you always get what you started with.

Pat


>--Frank
>
>pat hayes wrote:
> >
> > I entirely agree with Dan here. I just wanted to have an informative
> > (but stable) reference, not to imply any kind of endorsement or
> > normativity. I only use Ntriples as a way to describe graphs, since
> > its quicker to type than to draw a picture.
> >
> > We could eliminate Ntriples from the MT document altogether and give
> > all the examples as explicit graph pictures, like the M&S does.  I'd
> > be willing to do this, if the WG feels it would make the document
> > more accessible. I guess I was assuming that we were going to
> > 'publish' Ntriples in any case, even if it wasnt somehow endorsed as
> > a standard, but if that assumption is overly naive then the JPEG
> > alternative might be preferable.
> >
> > I await guidance.
> >
> > Pat
> >
> > >Art Barstow wrote:
> > >[...]
> > > > In the Open Issues of [2]:
> > > >
> > > >  http://www.w3.org/2001/08/rdf-test/#intro_issues
> > > >
> > > > The issue "is N-Triples normative?" is raised.  Given the
> > > > MT's reference to N-Triples it seems the answer is yes.
> > > >
> > > > What do the WG members think about this issue?
> > >
> > >I don't really know how to answer questions of the
> > >form "is X normative"? Here are some questions
> > >I can answer that are perhaps relevant:
> > >
> > >       1. does our specification of n-triples
> > >       constrain the set of XML documents that
> > >       conform to the RDF 1.0 spec?
> > >
> > >       No.
> > >
> > >       2. does our specification of n-triples
> > >       constrain software that conforms
> > >       to the RDF 1.0 spec?
> > >
> > >       No.
> > >
> > >       3. does the RDF 1.0 spec say that RDF/xml
> > >       is the only syntax in which RDF may be expressed?
> > >
> > >       No.
> > >
> > >       4. does/should the model theory *depend* on
> > >       the definition of n-triples?
> > >
> > >       No; it's independent of any particular
> > >       serialization of RDF.
> > >
> > >       5. is it useful to use n-triples as
> > >       an editorial device in the model theory document?
> > >
> > >       Yes, I think so.
> > >
> > >       6. should the model theory draft cite the n-triples
> > >       definition from the testing draft?
> > >
> > >       Yes; given 4 above, I suppose this should
> > >       be an informative citation, if we're going
> > >       to distinguish informative from normative
> > >       citations at this point in the drafting process.
> > >
> > >
> > > >  [1] http://www.w3.org/2001/sw/RDFCore/ntriples/
> > > >  [2] http://www.w3.org/2001/08/rdf-test/
> > >
> > >--
> > >Dan Connolly, W3C http://www.w3.org/People/Connolly/
> >
> > ---------------------------------------------------------------------
> > (650)859 6569 w
> > (650)494 3973 h (until September)
> > phayes@ai.uwf.edu
> > http://www.coginst.uwf.edu/~phayes
>
>--
>Frank Manola                   The MITRE Corporation
>202 Burlington Road, MS A345   Bedford, MA 01730-1420
>mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-8752

---------------------------------------------------------------------
(650)859 6569 w
(650)494 3973 h (until September)
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Thursday, 30 August 2001 17:59:54 UTC