SUO: Re: Let's focus
On 4/8/02 22:42, "jim.s3@juno.com" <jim.s3@juno.com> wrote:
> [snip]
> List members: Let's cut way back on discussions that do not directly
> relate to the work of this group. The Ontology list is set up for that
> purpose (see suo.ieee.org for how to subscribe). It would be better to
> have a silent list (with hope for future progress) than an off-topic list
> that drives away those who could potentially contribute. If such
> discussions continue, I'll send some friendly reminders.
Thanks, Jim..
But I don't think this gets to the heart of the matter. We have three
wildly different proposals now under the aegis of SUO.
* There is SUMO, which is an attempt to capture (IMHO) a middle-level
ontology including natural and non-natural kinds. It's built using
a logic that is fairly well understood (formal semantics are lacking
so far)
* There is IFF, which is a purely mathematical exercise (by this I'm
assuming there are no strong philosophical assumptions built in, but
I admit I am least familiar with it), but promises to be useful if it
can be shown work in organizing a mid-to-upper level ontology(ies)
* Finally there is OpenCyc, which is fancy and huge, but to date there is
no proof or model theory for it. Thus, to use it, one has to accept
its inference engine as a "pig in a poke".
But calling for order now and asking members to focus on the existing
proposals may be like closing the barn door after the horses have left.
Perhaps what should have happened is some more disciplined discussion early
on on what the relevant organizational and engineering principles are, with
the goal of arriving at principles that could be agreed upon as a starting
point, after which construction of competing upper-level ontologies could
begin, and rational comparisons made. So far, the only person that has said
much about these kinds of organizing principles is John Sowa. Of course my
memory could be faulty on this, and I'm willing to be corrected, but it sure
looks this way.
And of course there is the issue of whether one single standard for content,
rather than structure based on a small number of matematical/philosophical
principles, makes any sense at all.
I don't want to sound too gloomy about this, but it seems to me that one of
the reasons that participation has flagged is that many group members do not
see progress toward a unified engineering goal being made (this is not to be
equated with lines of code or concepts/classes/types/properties being
introduced) I fall into that category. The W3C work on RDFS and the
DAML/Oil stuff, while not a very good example of Ontology IMHO, does seem to
have this discipline and is thus making progress toward a workable standard.
This of course poses a danger to any real ontology work having an
appreciable influence in the long run. Like I said before, if we don't come
up with something really good and fast, we will all be forced to sit an
watch while the term "ontology" gets redefined extensionally by the W3C to
be those things that the W3C does.
As much as I get frustrated with the often obstructivist tone of guys like
Awbrey, Delatre, and others, I fear that there may be a kernel of truth in
all of that text they generate. Not that I necessarily agree with what they
have to say about philosophy or logic, but I am afraid I have to agree that
the SUO initiative in the direction it is currently moving is likely not to
terminate in success.
Sorry, no solutions today. Just some observations.
.bill