Re: What is an RDF Query?

From: Sandro Hawke <sandro@w3.org>
Subject: Re: What is an RDF Query? 
Date: Tue, 11 Sep 2001 22:34:22 -0400

> 
> Perhaps the confusion is in calling RDF a language.   I don't care for
> the word "Resource" in the acronym, but "Description Framework" seems
> pretty good.   

I argue against lots of things in RDF, but I certainly don't argue against
it being a language.  

> RDF says: when you want to convey information, do it in <object,
> property [relation], value> triples.  And it gives an XML syntax for
> those triples, and some more details.
> 
> And it provides some guidance (which may do more harm than good) about
> what symbols one should use in those triples to denote certain
> important objects, like the 'type' property, or web pages.

The problem with this part of RDF is that this part of the language has
been insufficiently specified.

> But maybe calling that a "language" is overblown.  To have a language,
> you really need to define what objects are being represented by your
> symbols, I guess, and RDF sure doesn't do that.  The p[e]ople using RDF
> for particular applications do that -- but in doing so they are
> defining *in english* what their symbols mean.  And therefor[e] making
> some language (probably with informal semantics).

Ah, what you want from a language is some intended meaning, i.e., something
like ``the real meaning of the URI <http://pat-hayes.com> is that logician,
you know the one I mean---the one that wrote the Naive Physics Manifesto
all those years ago''.   

> But because they are using RDF syntax, their language and someone
> else's language can be combined meaningfully.

How can this be done?  Do you mean by people?  Do you mean by some formal
reasoning system?  Do you mean by applications?

> I wish I knew a good way to for us to be clear about this with each
> other.   There may well be real, fundamental problems with the system,
> but damn if it isn't hard to tell amongst all our linguistic
> confusion.

Too true.  I am arguing that RDF, and uses of RDF, like using RDF in
more-powerful systems should be analyzed using the same metrics
that are applied to formal representation systems, such as first-order
logic or Montague logic.  If you have some other set of metrics, then I
think that you need to say what they are, and, moreover, to have them
subject to the same level of scrutiny that has been applied to existing
metrics for formal systems.

> > RDF IS NOT A PROGRAMMING LANGUAGE!  ASSERTIONAL DESCRIPTIVE LANGUAGES 
> > ARE FUNDAMENTALLY DIFFERENT FROM PROGRAMMING LANGUAGES!!  THEIR 
> > SEMANTICS DON'T WORK THE SAME WAY!!!
> 
> Sorry, I understand the difference between imperative and declarative
> systems, but I don't see why that's relevant in discussing
> modularization.  In particular, the appeal to C, which is of course a
> very different language, was to the idea that external modules can be
> written in a more expressive language (eg one capable of doing I/O).
> (In fact in C that's done via the asm language construct, but normal
> programmers never use that; it's hidden in the library.)

I think that you need to, at least, do some homework here.  Consider, for
example, the issue of universality.  Turing machines are generally
considered to be universal in some very strong sense.  However there is a
very different sense of universality for logics, where there are many
logics whose inference relationship cannot be implemented by a Turing
machine.
> 
>    -- sandro

Peter F. Patel-Schneider
Bell Labs Research

Received on Tuesday, 11 September 2001 23:13:41 UTC