W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2000

(unknown charset) Re: SV: The semantic web

From: (unknown charset) Dan Brickley <danbri@w3.org>
Date: Tue, 11 Apr 2000 10:16:17 -0400 (EDT)
To: (unknown charset) www-rdf-interest@w3.org
Message-ID: <Pine.LNX.4.20.0004110855240.19055-100000@tux.w3.org>


To add to what Tim said [1], a few points in response to that paper and
subsequent discussion...:

In that paper, RDF (including one XML syntax for RDF) is compared point
for point with other systems (OntoBroker, SHOE) that adopted a different
partitioning of the problem space. RDF Model, Syntax, Schema 1.0 doesn't
have notion of inference rules. SHOE and OntoBroker do. We called it RDF
because it is a _framework_ for describing resources -- rather than
attempting to do everything in one monolithic spec, RDF adopts a layered
approach. In hindsight the layering might have been different
(eg. different specs for
Model and for Syntax), but I believe the basic strategy remains sound.
Inference/logic comes later, as and when we can show it to be a useful and
well understood candidate for Web standardisation. In meantime, we're free
to use these tools with RDF today and to discuss here the feasibility of
scaling these technologies to the Web.

Think of RDF as inference fodder for smarter systems to consume; rather
than picking the one true inference/rules/logic language for the Web, the 
core RDF specs create an environment or marketplace in which these systems
can compete. 

As Sergey notes [2], many languages can be built on top of the RDF
core. If they prove useful, they may find their way onto the W3C RDF /
Semantic Web work. Just as the original Web provided a loose conceptual
umbrella over disparate information systems (gopher, ftp, http, wais), the
Semantic Web should provide an overarching framework for logic and agent
systems to compete (both in the standards-track sense and in the
electronic marketplace sense). At some point (perhaps as a result of
efforts such as DARPA's DAML programme) some leaders may emerge from this
marketplace and go through the W3C process; regardless, RDF was designed
to provide structured data for these systems to use and exchange.


Regarding the syntactic expression of RDF, the OntoBroker folks have made
some very sensible points about data duplication, and the need to draw on data
implicit in (typed) hyperlinks, semantics of tags such as <html:title>
etc.

A couple of things to say on this: firstly that a new, better, more
compact, intuitive etc etc RDF syntax only goes part-way to addressing the
syntax issue. RDF distinguishes between model and syntax for a good
reason: we want to be able to take HTML typed links, XML Links, annotated
DTDs, XML Schemas, Schematron, RELAX etc schemas and map them into a
common information model. I believe the fact that we have "the" RDF syntax
in the current recommendation has tended to obscure that goal. While
having another convenient RDF syntax also seems desirable, the important
thing is the model. Secondly, recent discussion here on "Semantic Web
Screenscraping"[3] suggests a way forward on the syntax front: we can
extract RDF models from arbitrary web content with XSLT. In the future we
may have other strategies for doing this, in particular XML Schema
annotations are a likely mechanism; these would allow instance data to be
less verbose since the mapping into the RDF model would be facilitiated by
an understanding of the syntactic schema being used.

does this make sense? 

Dan


[1] http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0113.html
[2] http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0119.html
[3] http://lists.w3.org/Archives/Public/www-rdf-interest/2000Apr/0024.html

On Tue, 11 Apr 2000, Greg FitzPatrick wrote:

> If you look in the mail archives you will find Tim B-L's comment on this.
> 
> You will have to look carefully since Tim misspelled the names of the
> authors.
> 
> Which just goes to show that the semantic web will have to be "spell-proof".
> 
> Greg
> 
> > -----Ursprungligt meddelande-----
> > Från: www-rdf-interest-request@w3.org
> > [mailto:www-rdf-interest-request@w3.org]För Tom Van Eetvelde
> > Skickat: den 11 april 2000 14:10
> > Till: www-rdf-interest@w3.org
> > Ämne: The semantic web
> >
> >
> > Hello RDF community,
> >
> > I have read the article "Practical Knowledge Representation for the Web" (
> > http://www.cs.vu.nl/~frankh/postscript/IJCAI99-III.html#Calvanese:
> > 98AAAI ) and to my surprise, RDF
> > is depicted as an immature and disappointing language for
> > representing semantics. This is in clear
> > contrast with the promise of RDF as the language to make the
> > Semantic Web come true.
> > The authors could be biased as they come from the AI world, but
> > nevertheless, their arguments seem
> > well founded to me.
> >
> > Does anyone have comments on this article?
> >
> > Regards,
> >
> > Tom.
> >
> 
Received on Tuesday, 11 April 2000 10:16:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:43 GMT