W3C home > Mailing lists > Public > semantic-web@w3.org > October 2007

Re: On Defensiveness (was Re: RDF: XULing or Grueling)

From: Harry Halpin <hhalpin@ibiblio.org>
Date: Mon, 08 Oct 2007 07:58:07 -0400
Message-ID: <470A1B4F.7080903@ibiblio.org>
To: Bijan Parsia <bparsia@cs.man.ac.uk>
Cc: 'Semantic Web' <semantic-web@w3.org>

Bijan Parsia wrote:
> I'm splitting this into two replies because I think it's good to
> separate out these issues.
> On Oct 6, 2007, at 12:01 PM, Harry Halpin wrote:
>> Bijan Parsia wrote:
> [snip]
>>> The linked data stuff seems harmless in that afaict it doesn't hork
>>> anyone off and seems sorta neat (though I've personally not read a lot
>>> of excitement about it from outside our community; pointers are
>>> welcome).
>> I mean, one could probably say the same thing for OWL 1.1 if one wasn't
>> deeply involved in the effort :) Let's try to keep positive here...
> [snip]
> First, it's really hard to see how I could have been *more* positive.
> I worked hard in this post and in particular in this section to *not*
> be unduely harsh. I am very clear that it is *me*, who at the start of
> the post set himself up as perhaps not being fully engaged and seeking
> to be brought back up to speed, who have not read about the
> excitement. I *am* reading the criticisms that prompted the post.
I just thought "harmless" was a bit off-putting - again, the WWW2007 BOF
session on Linked Data was more well-attended and exciting than some
sessions and workshops, and Tom provided some nice pointers already.

In fact, the Open Linked Data project could be really important. Unlike
say XML or JSON, because RDF is URI-centric and can in theory take
advantage of the "follow your nose" principle in order to get the
network-effect for the value of data, the real lack of substantive and
interesting RDF data has led people who are interested in using RDF to
have a sort of "where's the beef?" reaction. Open Linked Data can fix this.

Again, perhaps it's the Webbish side of RDF that we need to work more on
to take advantage of, instead of regarding it as a weirdo for semantic
networks that just happens to use URIs and be serialized in XML.
> So the only point of this comment is that I cannot say ANYTHING that
> might *remotely* be construed by *ANYONE* as critical of the vast
> success of the linked data movement. Frankly, that pisses me off.
> (Note, everyone else has been supportive about my call to reflect
> critically. Thank you all.) 
I would have said "interesting" rather than "harmless" - or possibly
even "helpful" :) However, Bijan, you know we all respect your work on
RDF and OWL a lot and know you have done more than most people to deploy
them in both academic and commercial contexts. That's why there was a
":)" at the end of my comment. Not meant to piss anyone off.
> Second, this is a sheer *tu quoque* argument. Let me start out by
> *embracing* the thing one could probably say. It's not just that you
> can *problably* say it, you *can* say it! It's *true*! So what? What
> does that have to do with being worried about the fallout of past RDF
> showcases failing hard? I'm *glad* OWL 1.1 isn't generating a lot of
> excitement outside the community, and don't mind if it doesn't
> generate a lot of excitement within it. I think it's necessary for
> some times and helpful for others and probably will be overall worth
> the investment. But, fortunately, OWL doesn't seem to have these time
> bombs. It has different marketing problems.
Not "tu quoque", I was just pointing out the general critique of the RDF
data-model not being suitable for all problems could easily be applied
to OWL, XML - or anything really. I think we all know that RDF isn't the
solution to world hunger. I don't think many people are evangelizing it
as such. I tend to evangelize it as a data-merger solution, i.e. merge
data and query using SPARQL is useful and can solve certain specific
problems. See GRDDL Primer for examples [1].
> I would have hoped, Harry, that you wouldn't stoop to such fallacies.
> I tend to market RDF and OWL in very specific ways and, I think, have
> been reasonably successful on both fronts in a lot of cases. However,
> in both cases I try to be *very* clear when they are *not* a good fit,
> in my opinion. Or even where they'd work, and work well, but might not
> work well enough *better* than an existing system to be worth the
> transition costs.
> Cheers,
> Bijan.


Harry Halpin,  University of Edinburgh 
http://www.ibiblio.org/hhalpin 6B522426
Received on Monday, 8 October 2007 11:58:13 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:03 UTC