xml-dev - Re: local, global (was various ontology, RDF, topic maps)

["Simon St.Laurent" <simonstl@simonstl.com>]

At 09:13 AM 12/20/00 -0700, Uche Ogbuji wrote:
>I certainly agree with your viewpoint, but I can't identity any GUIM
>practitioners from the discussions we've been having.  I didn't meet all
>the folks you did at XML 2000, and it sounds as if I missed a few spicy
>sessions, but I think that most people with whom I've been seriously
>talking about ontologies are pragmatists.

Perhaps it's just the time I spent listening to various folks at XML 2000,
but I remain seriously concerned that what you identify pragmatic approach
leaves the door open for what I'd call the idealistic approach, with
potentially ugly consequences for the pragmatists.  Maybe there's room for
a pragmatism of limited expectations, but that's something I hear from only
rarely.

>I myself have mentioned many times that my observations apply to closed
>systems and that I don't claim any insight into how to make something like
>a semantic Web work.  My perception is that Len, Martin and Jonathan don't
>make much more radical claims.

I find it remarkable that people talk about 'agreement' as if it weren't a
radical concept in itself, and that they actually seem to believe that
maintaining large sets of agreements about meaning is both possible and
beneficial.  I'm heartened by discussion of the contingencies involved in
such projects, but still find the foundations far less solid than large
groups of people seem to believe.

>The record of authority is the key, as Len pointed out.  It pretty much
>defines the scope of the model.  I'm skeptical of a record of universal
>authority: I don't think even ISO BSR represents such ambition (note the
>"basic" in the name).

Are records of authority inherently self-limiting?  Or do they effectively
claim control over a given field?  I'm not sure that remapping
centrally-defined visions of information to the real world is really how
people can best apply XML, though it seems to be the model currently put
forth as orthodoxy.  

I'm well-aware that businesses and governments feel more comfortable with
'stable', 'controlled', 'authoritative' and 'credentialed' approaches, but
I worry that such faith in a conservative vision of information modeling is
badly misplaced, with potentially ugly consequences.

Relational database people seemed to think for years that information which
didn't fit their model wasn't important.  I'm worried that we're headed for
a similar though larger problem where information that doesn't fit a given
large-scale agreement is deemed unimportant.

>Just in case, I should mention that there is no GUIM in the idea of
>layering the Topic map model on RDF.

I don't think GUIM is necessarily inherent in either RDF or Topic Maps.
Unfortunately, they both seem to attract that crowd, and I'm concerned that
it will come to have a significant effect on best practices, if not the
specs themselves.

>I'm still quite interested in hearing the whole "semantic Web" story.  I'm
>sorry I missed the XML 2000 session because I'd like to learn what the
>true ambition is, and the practical developments required before it can be
>realized.

Edd Dumbill has a good summary on XML.com:
http://www.xml.com/pub/a/2000/12/xml2000/timbl.html

I also strongly recommend Tim Berners-Lee's book, "Weaving the Web".  It's
got some harder selling in it than his presentation did.

>I don't suppose there's any chance for someone to set up a panel discussion
>at some conference with TBL, Len, Martin, Howard Smith and a few other clever
>folk.  I'd hitchhike to go see it.

They're doing a Semantic Web day at WWW10 in Hong Kong next year, but I
don't know if it'll have your favorite speakers.  It'd be a hell of hitchhike!

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books