A certain difficulty from Tim Berners-Lee on 2000-02-29 (www-rdf-interest@w3.org from February 2000) (original) (raw)

-----Original Message----- From: David Megginson <david@megginson.com> To: xml-dev@xml.org <xml-dev@xml.org>; www-rdf-interest@w3.org <www-rdf-interest@w3.org> Date: Friday, February 25, 2000 6:07 AM Subject: Re: A certain difficulty

Pierre-Antoine CHAMPIN writes:

I would not say that either ! I find the RDF model very simple and uniform (it's all about triples) which makes its elegance... and for some people its weakness !

Unfortunately, it's not about triples. The only way to discover the true RDF data model is to reverse-engineer it from the XML, and it turns out that there are at least six components (not three) in each statement:

subject subjectType (global id, local id, URI pattern)

The global ID and local ID are IDs, so the semantics should be the same.

A reference with a #fragmentID is taken as a reference to a part of (or view of) a document, while a reference without a # is taken as a reference to the document.There is an unspoken problem that in the RDF spec a reference to a subtree of an XML document containing RDF is taken to be a erference to the RDF object. .........(1)

The URI-pattern I agree is a big problem, and I think a lot of people would have wished it not there. I wonder how many systems implement it. It seems to me best to put it off to a level of logic above the basic RDF. .............. (2)

predicate

object objectType (literal text, literal XML markup, reference)

The object being the union of litteral types and reference to node is reasonable: the object may be represented as a pair (type, value) for example (or some other syntax or a pointer into a different part of memory or a pointer to a self-typed object or whatever.)

You could argue (and people have i understand) that the same ought to hold for the subject of course. ............... (3)

objectLang

This is a mess pure and simpe: it is in the syntax and not in the model. How did that happen? The syntax should not have bowed to the internatoinalization community with a syntactic nod, but instead asked for an RDF vocabulary for language. It should be removed from the synatax. ..............(4) My nurdlings on two ways of doing it are in http://www.w3.org/DesignIssues/InterpretationProperties

These are not simply syntactic artifacts -- it's information that must be exposed through any RDF API, and thus, part of the core model, independent of the peculiarities of the XML markup (note that I'm assuming that bagID, etc. are predigested). The URI patterns (aboutEachPrefix), especially, make it much trickier to do any relational database implementation of RDF, since you the set of possible subjects is open.

yes, abouteachprefix was a mistake in retrospect.

Here are 4 issues for an RDF M&S retrospecive post-Rec issues list! Ralph or Dan Brickley was going to draw one up I think.

In the contrary, the XML syntax is a bit confuse, true.

Yes, it is also unnecessarily confusing.

In my point of view, the problem comes from the recommandation mixing modeling and syntaxic aspects (I won't mention semantic aspects !) in a way it's hard to differentiate them without some RDF experience.

The problem is that the model as presented is naively simple, and the WG failed to notice that the XML syntax is not based directly on that simple model.

All the best,

David

-- David Megginson david@megginson.com http://www.megginson.com/

Received on Tuesday, 29 February 2000 17:08:02 UTC