Re: Why is RDF such a tough sell?

On Mon, 24 Jun 2002, Stephen K. Rhoads wrote:

>Why is RDF such a tough sell?  Two reasons, IMO: [...]

I'd add the classical one, that is, a killer application which hasn't got
anything to do with XSchema. Right now, the most succesful RDF (or
RDF-like, in some versions) application is RSS, which is indeed handled
more like a document type than an instance of RDF.

>Imagine the search engines of the future; they can either be built to
>understand *just* the RDF model and "plug in" vocabularies as needed, or
>they will be stuck attempting to understand potentially thousands of
>independently developed XML Schemas.

I'd add syndication and metadata harvesting, here. In the RDF framework,
neither of these applications needs to know anything about what is
processed, but they can still function. They become pure middleware, using
a common vocabulary expressed by RDF. That's a lot cleaner than having
each intermediary component understand the data it's passing through. This
sort of thing might sound like a pure engineering benefit, but it's real
enough if syndicated dataflows ever grow large enough.

>I suspect that the human development time will be orders of magnitude
>greater for the latter option.

Quite. I think this might well be the best selling point of RDF once
sufficient component support becomes available so that Joe Average
Programmer starts to think of RDF as the the easy way out.

>I am reasonably technically proficient and yet I had to read (a) the
>Model/Syntax Specification, (b) the RDF Schema Specification, (c) the
>RDF Primer, and (d) various peripheral documents TWO OR THREE times each
>before I "got it".  And yet, RDF is a very simple, yet powerful,
>protocol.

Quite. What we need is documentation your grandmother can understand,
templates for the most common RDF Schemas, semi-automatic online
generators and easy, locally usable whatever-to-RDF converters.

>3. It's not yet deployed in HTML editors/browsers.  Once folks start
>plugging metadata into their run-of-the-mill HTML editors and see RDF
>statements popping up in their code, the landscape will change.

I'm thinking, those who have used RDF for longer should really list the
most common applications where they think RDF will have an impact. If
those applications get useful model implementations, either as browser
plugins, browser infrastructure proper or server side applications, one
would think that RDF could gain a lot of strength.

The first three on my list would be RSS done in RDF proper, sitemaps and
FOAF presentation/its integration into existing phonebook architectures.
The first would enable private RSS extensions which many sites are likely
to want once they know they're interoperable, the second is a common
enough problem to warrant a standard, no-fuss implementation, the third
solves the common problem of people not being able to easily use and
exchange each other's personal information (i.e. the kinds of electronic
business cards you get with cell phones (vcard) do not yet exist in the
online universe; one would think a few standard FOAF apps could solve this
one).

Sampo Syreeni, aka decoy - mailto:decoy@iki.fi, tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2

Received on Tuesday, 25 June 2002 04:33:07 UTC