W3C home > Mailing lists > Public > public-owl-dev@w3.org > January to March 2008

Re: post-structuralistm and (formal) ontologies

From: Danny Ayers <danny.ayers@gmail.com>
Date: Mon, 4 Feb 2008 19:51:53 +0100
Message-ID: <1f2ed5cd0802041051g596a2an626fe6b479e5b99c@mail.gmail.com>
To: kendall@clarkparsia.com
Cc: "Owl Dev" <public-owl-dev@w3.org>, "Jeremy Carroll" <jjc@hpl.hp.com>, "Karl Dubost" <karl@w3.org>, "Dan Brickley" <danbri@danbri.org>

On 04/02/2008, Kendall Clark <kendall@clarkparsia.com> wrote:

> Unless one can show that OWL or RDF are *especially* prone to or
> implicated in some kind of oppressive social structure, this line of
> inquiry *qua an OWL or SemWeb issue* just isn't very relevant in this
> context. And Dan's said enough, IMO, to suggest a reasonable prima facie
> case that neither OWL nor RDF nor any other appropriate topic for this
> mailing list are *especially* troublesome w/r/t Jeremy's inquiry.

Presumably a significant part of the motivation behind ideas like
RDF/OWL is that we can do stuff with them on computers, so in one
sense given the nature of computers as currently deployed it's hard
not to agree - gross approximations are a feature, all the way down.
0s are from Venus, 1s are from Mars (or vice versa).

However a system encoding "now = day | night" would leave few any
doubt over it being an oversimplification when it came to describing
the real world. But as has already been pointed out, with more
sophisticated languages it is possible to make questionable statements
appear reasonable/reasoned.

Possibly more worrying still is the projection of this, the way in
which the system is interfaced to the end user. The "friend or foe"
systems on national borders are shifting to increased software
involvement. The prejudice would be pretty visible if a person was
denied entry to a Western country on the basis of a border guard not
liking the look of their turban. But it's not hard to imagine similar
prejudices being held in the machine's knowledgebase. Garbage in,
garbage out applies to T-boxes too. A good case for timbl's "Oh yeah?"
button I suppose.

I know it sounds like the "guns don't kill people" argument, but it
does seem like the application of the technology matters. When we're
aiming for tech with a broad range of application, that becomes

"now = day | night" is probably a good enough approximation for an
automatic patio light, a system unlikely to have political side
effects. But many of the systems using RDF/OWL do have political side
effects, so in that sense the issue may be more troublesome.

> Thus, while I think Jeremy's line of inquiry is interesting enough to
> pursue, and I did actually do some work on it back in 2000, it's not
> something that computer scientists, qua computer scientists, are
> especially well prepared to work on. It's a question of the politics of
> technology or social informatics more broadly construed, and should be
> approached w/ the tools, methods, and background knowledge of the
> practitioners of *those* disciplines.

In general I agree, though do wonder if computer scientists might have
insights on the kind of obfuscation-by-sophistication mentioned above,
in that they may be in a better position to compare what's on the
surface with what's under the hood.


Received on Monday, 4 February 2008 18:52:03 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:58:16 UTC