[topicmapmail] Summary to date of "context" and "bottom up"

[Jack Park]

From: Bernard VATANT <b.vatant@wanadoo.fr>


 <JP>
 *We* is a big word.  I imagine that *we* will be several groups
 (communities, etc)--a set of communities as Martin Bryan pointed out. I
 further imagine that this means that any particular *public subject* will
 cover more than one arrangement of topics, according to the various
contexts
 that emerge.

 Yes, I  say we need convergence.  But, the *we* can be many groups as just
 mentioned, and each individual *we* should have some purpose or, if I
recall
 my 40 year old French, un raison d'etre (I've waited a long time to try
that
 ;-).  A purpose, to me, implies something other than continuous mental
 gymnastics.  I also imagine that convergence will occur about a moving
 target, an attractor basin, rather than necessarily a fixed point.  Of
 course, in any ontology, there are going to be many fixed points, complete
 agreement by all communities on some notions. It seems to me that
attractor
 basins will likely be far more interesting, as, for instance, any response
 to the question "What do you mean by *we*?"
 <JP>

<BV>

I set this question - who are "we" ? - not to get any answer or definition,
which I agree is not very interesting in itself, but to invite to some
prudence in the using of that collective term, and before using it make
sure the context gives sense to it, and if it does not, replace it by
anything more explicite.
Anyway, about convergence. My old-fashioned maths education leads me to
give a default sense to that term, (that there is a unique limit). That's
why I answered about unique thought. Sorry about that. If you think in
terms of attractors, I begin to understand what you're up to, and am ready
to explore with you that interesting track. Maybe - I throw it there just
as it comes right now out of my mind -  we can somehow identify a "context"
to such a semantic attractor. It may lead to some "objective" approach of
what a context is, considering it more like the result of a process than a
set of definitions coming out of any fuzzy metaphysical or ideological a
priori ;-)
(BTW, who is Gordon Pask ?)
Any idea about the kind of "semantic metric" in which define the
convergence process and discover such attractors - if any ?

<JP> A purpose, to me, implies something other than continuous mental
 gymnastics </JP>

I wonder who'll take that one :(  ... Hope the project which brought me
here could be considered like a valid purpose ... *une* raison d'être !

</BV>

<JP>
Gordon Pask created something called Conversation Theory.  Google that and
be prepared for a surprising amount of links.  *une* Hah! Now, I remember
why I did not do all that well in French classes.  Thanks.  I'll work on
that. I'd like to think I'll eventually be able to read your web pages in
French :-)

Let me make it very clear that I rarely think of convergence as *requiring*
a limit or fixed point. Life's justs too chaotic for me to want to go there.
I like to think in terms of tendencies; a context represents, at least for
me, a tendency -- an attractor basin filled will all sorts of notions and
lots of people orbiting about those notions.

Now, how would one imagine implementing attractor basins?  I imagine it this
way: suppose we take two or more ontologies and we compare them.
Individually, each may contain representations that are the same or similar
to some or all of the others.  Out of that, there may exist a kind of
intersection that represents a kind of tendency.  It may well be that
statistical thinking will be needed here. Indeed, there exists work on the
combination of Bayesian math and Description Logics at Stanford and
elsewhere.  Each individual ontology could be viewed as an *image* of the
intersection, so, perhaps the intersection reflects a kind of tendency that
describes the attractor basin formed by the images.  Given that we humans as
individuals almost never *see* things (whatever that means) precisely the
same way as others, it may be that some attractor basins could conceivably
include an enormous number of individual ontologies.  What's attractive
about that is that it opens the possibility for enhanced interaction given
the opportunity to *see* some concept from another point of view.

Semantic metric?  How about something like average percentage of individual
ontologies covered by the *tendency* ontology? Like everything else in the
universe, that value would evolve over time.  Nonlinearity brings attractor
basins into existence. I cannot imagine a constant metric in a nonlinear
world.
<note>My view on images is not mine as an original thought. Indeed, I have
probably never had an original thought in my life, but I must point out that
I began to see the notion of images more clearly while listening in on a
conversation between Howard Liu, Donald McKay and Leo Obrst, all ontologists
here at VerticalNet. Plus which, I'm not sure I have the concept of *image*
right.
</note>

About convergence, I believe that this and other conversations are
contributing to an evolving *image* (mine), and that image gives hints of
converging on some solid ideas.
<JP>



============================================================================
This message is intended only for the use of the Addressee(s) and may
contain information that is PRIVILEGED and CONFIDENTIAL. If you are not
the intended recipient, dissemination of this communication is prohibited.
If you have received this communication in error, please erase all copies
of the message and its attachments and notify postmaster@verticalnet.com
immediately.
============================================================================