[topicmapmail] Re: contexts, attractors, authors and users

[Jack Park]

Nicely done, Bernard.  Just got back from London and the XTM working group.
Got to think about this quite a bit (18 hours in the air, to be exact).  I
wonder *what's in a topic*?  I have this notion that there must exist a
semantic attractor, one which exists somewhere in a space defined by an
author and a reader.  I wonder what that means.

Reference to Principia Cybernetica http://pespmc1.vub.ac.be/, and of course
to Francis Heylighen's seminal paper on Gordon Pask's work really brings my
thoughts into better focus.  Pask spoke of entailment meshes.  That fit
closely with another thread I have been following (Robert Rosen, _Life
Itself_ -- visit http://views.vcu.edu/complex/ for more) in which entailment
is a key notion.  A colleague here at work is studying, from a category
theoretic point of view, the ways in which individual ontologies can be
viewed together.  My naive thought is that if a group of ontologies exists
about a particular topic (various authors and readers), then that group can
be considered as a category itself for the chosen topic.  That way, it may
be possible to look for an algebra that ties the separate ontologies
together, quite possibly with some experiential (probabilistic) central
tendency emerging. The work of Geo Wiederhold in the algebra area is on my
short list to examine. Probabilistic issues remain hot, but somewhat
unstudied at this moment.

For me, level of importance shifts according to context.  I am thinking that
a student reader is liable to bring less semantic baggage to any read of an
ontology authored by another individual, while an experienced reader may,
indeed, see things differently, quite possibly contributing to a dialog that
enhances the central tendency of the category.

>From the perspective of watching browsing patterns, Heylighen and Joslyn
have spoken at Principia Cybernetica of a kind of Hebbian learning
associated with web links. Harry Klopf, in unrelated work, has managed to
take Sutton and Barto work to a new level called differential hebbian
learning, and there may be some value in looking at that work.  In any case,
hebbian learning begins to relate to the *experiential* or probabilistic
learning referred to above.  But, in my naive view, just following links
does not make them either more valueable or reliable.  Something else is
needed.

I do not believe that topic maps need to be coherent in the ontological
sense to be of great value in the community process of sharing ideas and
building ontologies. Topic Maps follow a long history of work in
constructivist learning research that resulted on Concept Maps, Mind Maps,
and so forth.  What is clear, at least to me, is that topic maps hold a
valuable place in the evolution of knowledge, even if for serving no other
purpose than indexing.  For me, this discussion is about pushing TMs beyond
indexing.

I believe that a larger participation in discussions such as this will
enable the emergence of a new understanding of knowledge, its growth and
management.  I also believe that, to make that happen, a new kind of
collaborative tool will be needed, one that promotes the online construction
of real and useful ontologies as discussion occurs.  Part of that work can
be automated, and part of it will require human dialog.  Following
inspiration of Douglas Engelbart (http://www.bootstrap/org) and his OHS/DKR
project, I have begun to build sketches of software that will support such a
web site.

Jack Park

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

> The debate over contexts seems quite in standby ... I'd like to merge two
> ideas posted lately.
> Martin pointed that in definition of a context, user viewpoint is as
> important as author's (even more)
> Jack "converged" to the idea that context is something like a semantic
> attractor.
> What shall we do with that ?
>
> If we think of ontologies in terms of nodes and links (or topics and
> associations), the user viewpoint will be the way he browses the
> associations map. Let's take a set of topic maps, linked through
> associations, but not necessarily coherent from an ontologic viewpoint,
but
> a priori relevant to a community of users. Let users browse this set of
> maps freely, following or not proposed associations. It should be
> interesting to analyse this browsing in terms of trajectories in the
graph.
> The result of such an analysis should lead to attractors, from which
> authors could refine their previous conception.
>
> Such exchange between authors and users - in fact involving of users in
the
> authoring process, even if they don't know it - seems a way to make
> converge authors' and users' ontology, and hence defining contexts, not
> from dogmatic viewpoint, but pragmatic one.
>
> Does anybody knows about precise works in that direction, other than too
> much confidential-strategic information ? I found some 3 years old work at
> http://pespmc1.vub.ac.be/Papers/BootstrappingPask.html
> dealing with that sort of idea, but not with the topic maps standard.
>
> Bernard VATANT
> b.vatant@wanadoo.fr
> www.universimmedia.com
>


============================================================================
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.
============================================================================