W3C home > Mailing lists > Public > semantic-web@w3.org > August 2006

Re: Is there real world RDF-S/OWL instance data?

From: Danny Ayers <danny.ayers@gmail.com>
Date: Sat, 5 Aug 2006 11:09:35 +0200
Message-ID: <1f2ed5cd0608050209p5766db07u72fbee6b7c2b630d@mail.gmail.com>
To: bob@snee.com
Cc: "Frank Manola" <fmanola@acm.org>, "Sören Auer" <auer@informatik.uni-leipzig.de>, semantic-web@w3.org

On 8/4/06, Bob DuCharme <bob@snee.com> wrote:
> I tend to alternate between optimism like Harry's and pessimism.

To me a year or two to realise the Semantic Web wasn't going to happen
overnight, but I've pretty much settled on Qualified Optimism : a Web
of Data is inevitable IMHO, but whether the primary language for that
data is RDF is another matter (bearing in mind that XML is also part
of the SemWeb stack). But RDF is the best we've got right now for
generalised cross-domain interop, so it's well worth the attention.

> The incentives to create it just aren't there. We have to think of some.
> For example, if MovableType started putting RDF/A metadata in its default
> templates instead of commented-out RDF/XML, that would be one jump start
> for getting a lot of parsable triples on the web. What other kinds of jump
> starts are we still waiting for?

I don't think we can talk of incentives in such immediate terms just
yet, what in this case would be MovableType's incentive? To create
incentives there needs to be compelling demonstrations of what can be
done with RDF. This is almost back to a chicken & egg problem state.
But not quite.

There is a route to adoption through refactoring the initial question
- instead of "Is there real world RDF-S/OWL instance data?" try "Is
there real world data...how can it be addressed as RDF/OWL?

There is lots of data in relatively easy reach - thanks to SQL DB
adapters, GRDDL/XSLT on microformats and various XML languages etc.
SemWeb technologies can be usefully applied to the (transformed) data
- data integration with unified querying (SPARQL) is likely to be of
huge benefit. SemWeb tools are by their nature domain-independent, so
kit designed for one scenario could be deployed virtually unchanged in
another. (My personal favourite on this front is SIMILE's Longwell
facetted browser).

Going back to the Web of Data viewpoint, I reckon the Web 2.0
developments indicate a path ahead for the public Web - after all the
Semantic Web is the ultimate mashup. I also think the work around
syndication is likely to start showing real dividends in SemWeb terms
in the very near future.

There was always something of a conflict between publishing content
("simple" RSS), and publishing data (RSS/RDF). Now Atom has come along
as a clean way of publishing content, clean enough to potentially
support generic data (recent suggestions [1] have included RDFa
payload). The Atom Publishing Protocol implies an Atom Store as a
generic content store.  It's also Web-(HTTP/REST)-friendly. This
genericity in the domain of human-readable content seems to me like a
useful stepping-stone between the HTML Web and the Semantic Web.
(I've been referring to this general idea as the content model layer -
must do a proper write-up soon...).

Tools are being very actively developed to support this kind of
infrastructure. On the one hand the content-oriented developers are
approaching a good position to exploit SemWeb tools (a SPARQL endpoint
on an Atom Store would be very useful for alternate presentation), on
the other the cleaner content management tools that emerge can be
usefully piggybacked by data-oriented developers simply by switching
the payload.

So although I agree it's disappointing there isn't more raw RDF
instance data out in the wild, I think we're beyond that being a
blocking issue. The Web at larg hasn't stood still, and the way it's
moving is generally SemWebwards.


[1] http://torrez.us/archives/2006/08/02/476/

Received on Saturday, 5 August 2006 09:09:46 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:41:00 UTC