Re: RDF use case: Extending and Querying RSS channels

On Wed, 31 Jan 2001, Graham Klyne wrote:

> At 09:01 PM 1/30/01 +0000, Dan Brickley wrote:
> >context:
> >Someone (who shall remain nameless) just told me RDF Site Summary([1])
> >was all about ("mere") headline syndication, not meaty interesting
> >Semantic Web
> >inferency stuff. So I'm reminded to circulate this little note which
> >describes the use of RSS as a pretty generic transport for
> >application-specific Web data. Our example uses job descriptions, but
> >I've come to think the same trick works for a suprisingly wide range of apps,
> >eg shared bookmark feeds ([2]). The point of using RDF for RSS was that
> >it allows us to do exactly this; mix application vocabularies within a
> >common environment. RSS from this perspective goes way beyond headline
> >syndication. For many apps, the RSS-specific constructs (rss:item,
> >rss:channel etc) fall away. The important RSS-ish thing becomes the
> >practice of sharing data this way, rather than the specific RDF vocab
> >defined by RSS...
>
> I'm tempted to say "isn't this obvious"?  (The use of RDF for sharing
> application data in a modular, extensible, open-ended fashion as a
> consequence of the general goals of RDF.)

It's obvious to us (a geek minority), perhaps. The nice thing about RSS
1.0 is that applications that don't give a stuff about Semantic Web,
namespaces, RDF or whatever can still render an RSS feed in one of those
funky little HTML boxes with a logo, blurb and title/link/description
bullet points. While it's been "obvious" to me for 3-4 years, it's
rather nice to see it happening for real...

RSS 1.0 creates a marketplace for exposing and consuming RDF-happy data
feeds, without having to wait for fancy RDF query and aggregation tools
to mature. The more additional metadata one decorates a bare RSS feed with,
the more people... buy the book... apply for the job... download the
file... whatever. Sophisticated consumer apps are rewarded; expressive
service advertisements are rewarded; downlevel aggregators aren't
excluded.

> Or am I missing something important (about RSS)?

Yes. People understand it! :-)

In fact (see http://www.w3.org/2000/08/w3c-synd/) you can deploy RSS on
a site simply through making more disciplined use of (X)HTML.

The core problem that RSS 1.0 addresses is that because it's so simple,
people started to use RSS for all sorts of things. By allowing
namespace'd extensions, we ensure that a generic RSS environment can
allow application semantics to play well with 'naive' RSS consumer apps.

 >
> Or maybe I just happen to have fallen into this way of thinking?  A
> work-in-progress example of something I am developing can be found in "An
> XML format for mail and other messages" at
> <http://public.research.mimesweeper.com/Messaging/draft-klyne-message-rfc822-xml-01c.txt>.
> This isn't immediately obviously an RDF design, but if you peek at appendix
> D...

Yes, I stumbled across that earlier today, considentally.

Not being an 'obviously RDF design' is a good thing. An RDF view of all
this data should be a low-cost job. Having an RDF view of XLinks being
a similar example; buying into the RDF data model needs to be made as
cheap as possible.

> I'll also note that the design of CC/PP follows a similar pattern, if I am
> understanding your point correctly.
>
> There are also some notes from an internal company presentation I gave
> recently at
> <http://public.research.mimesweeper.com/RDF/RDFMetadataForEndToEndContent.html>
>
> Are we talking about the same general approach here?


I think so

Dan

Received on Wednesday, 31 January 2001 13:31:18 UTC