[topicmapmail] Heredity from one to another topic

[Paul Prueitt]

(I am posting also to the small e-forum ontologystream

http://www.ontologystream.com/forums/index.htm

as this discussion may have a larger scope that "Topic Maps".

Andrew, Bob, Nan, Bjorn  and others, are you members of the OntologyStream
forum?  If not would you might joining so that I can reduce my cc list.
Send me a note since although I am moderator of this ontology forum, I have
not the knowledge as to how to see the membership.. <sigh> )



Tina,

I am not so agile in my understanding of topic maps.  However, I see in your
question (copied below) something that I have some experience with.  The
issue you raise is one of terminological variation as words (topic maps) are
placed first in one context and then into another.

This issue you raise is not merely a question for the topic maps standards
committee, but is also relevant to the failures that Autonomy's dynamic
reasoning engines have.  The DRE's are not substitute for a topic map, and a
topic map, as yet, does not have a process model that allows "topic map
arithmetic" in the general case.

Without the ability to add and subtract topic maps, we do not have a means
to add new themes into a model of a discussion's references, nor to remove
themes from the "focus" or scope once this theme is no longer being
supported by the intent of the "speaker".

Nan's B-2-B/B-2-C architecture requires such arithmetic.  A specific example
would be:  A customer logs onto the Summit Racing Equipment e-commerce site
and asks a question about if the engine that she has rebuilt requires a part
of a certain type, since the original engine needed this part.  The answer
is no, that this part is no longer necessary to the rebuilt engine.  The
"automated" discussion then needs to proceed without the topic of that part
being in the topic map that "controls" the discussion between her and the
ordering system.  Later, it is discovered (in the mind of the customer) that
an alternative approach to re-building her engine is preferred.  This new
information needs to be added to the controlling topic map.  Suppose now
that the occasion has the need for the topic of the part of a certain type
to be discussed.  Most expert systems would not handle this in an ad hoc
fashion (thus the occasion's needs are ignored and the customer ends up
ordering the wrong parts.) Clearly Autonomies dynamic reasoning engines
would not have a remote change of handling this transaction.  In fact, my
use of natural language only is communicative if the reader is quite aware
of the occasion of this (present) communicative context.

The occasion of a situation has a root in the temporal experience of time,
whereas the situation does not. This root must "point" to an non-addressable
subject - that is the glue that hopefully binds together and understanding
that (like later Wittgenstein) points to a reality beyond the words.

The problem is subtle - as to whether it is un-solvable we do not know. The
use of natural language within communities that have collective memory (what
ever that is), also sometimes fails to communicate about the occasion of a
situation - the pragmatic axis.

As Robert Shaw (at the Einstein Institute) has recently reminded me, the
notion of content can itself be enriched to include context in a situation
and context in an occasion of a situation.  The situation has some degree of
general setting and can be talked about in the abstract, i.e. can be
represented either in natural language or formal systems.  The occasion of a
situation has non-addressable aspects - such as aspects that are specific to
the occasion's uniqueness.

Comparative terminology science is a very small academic discipline that has
put some thought to the issue of transactional (from one language to another
language) aspects of words that are missing the means to properly represent
situation, and the more difficult case of occasion of a situation.  This is
related to the Whorph-Saphire hypothesis in translational theory (sorry - I
may have miss spelled one or both of these names).

A process model for knowledge representation (topic maps) may need to be
enriched with some of this discipline.

As for references to this work - there is very little that I can point to
other than

a section, Why do we need more than one ontology?,  in my second chapter of
a manuscript at:

http://www.bcngroup.org/area3/pprueitt/book.htm

Most of what I know about "comparative terminology science" has been learned
from communication with Dr. Fiona Citkin, who I have cced in this message.

Perhaps you might start a conversation with her.




-----Original Message-----
From: topicmapmail-admin@infoloom.com
[mailto:topicmapmail-admin@infoloom.com]On Behalf Of Tina Friis Bjerrum
Sent: Sunday, April 15, 2001 7:03 AM
To: topicmapmail@infoloom.com
Subject: [topicmapmail] Heredity from one to another topic


Hi everyone,

I'm a student of computational linguistics at Copenhagen Business School
and I'm currently working on a project involving Topic Maps.
I have some hierarchic systems of concepts (systems used in the field of
terminology) and I'm trying to apply them to topic maps.
Since topic maps are new to me, I thought you might be able to help me
with this one question.

Is it possible to have one topic inherit characteristics form another. I
don't mean name, occurrence or role, but metadata on the topic such as
birth place, birth town etc.(which is defined as 'metadata occurrence
role type' in the opera.xtm by Steve Pepper) ?

Hope someone can help me.

Thanks and best wishes,
Tina

_______________________________________________
topicmapmail mailing list
topicmapmail@infoloom.com
http://www.infoloom.com/mailman/listinfo/topicmapmail