Re: [topicmaps-comment] Notions have existence .....

[Ivan Uemlianin <ivan@jurakm.com>]

A brief tupennorth on this debate: abstract objects are useful whether
they're real or not; subjectIdentity elements are useful either way; beware
technical solutions to non-technical problems.

[This is not intended to be a response solely to SRN's post, even though it
looks that way]

Steven R. Newcomb:
> Maybe I've been
> brainwashed by 2,400 years of thinking based on Plato's
> ideas.  On the other hand, maybe Plato and I have
> simply had the same unshakeable intuition that every
> idea is necessarily unique, has identity, and exists
> regardless of whether anybody happens to be thinking
> about it at any given moment.  I think it would be
> awfully hard to explain the phenomenon of language --
> information interchange -- in the absence of such an
> intuition.

There *are* traditions which deny the reality of abstract objects (e.g.
Aristotle and onwards).  These traditions are well represented in current
philosophy of mind and language, not to mention the natural and social
sciences.  Explanations of the phenomena of language - e.g. in linguistics
and psychology - are getting on quite nicely without reference to the Forms.
Even if one regards the biologically deterministic cognitive science of the
Chomsky/Fodor tradition as a kind of crypto-Platonism, there's plenty of
mainstream work in philosophy (e.g. social externalism), linguistics (e.g.
HPSG) and psychology (e.g. connectionism, behaviourism,
social-constructivism) which is either agnostic on or hostile to the
Platonist view.

But ... surely it doesn't matter whether abstract objects are real or not.
It makes no difference for most work in mathematics, the natural and social
sciences and the humanities, so why should it matter with topic maps?  For
example, it's hard to think of any work on Beethoven's Fifth Symphony to
which it would make any difference whether the abstract object 'Beethoven's
Fifth Symphony' was real or whether it was a convenient conventional fiction
to facilitate discussion and research.  In any case the usefulness of
subjectIdentity elements for explicit indications of a topic's denotation is
obvious.

Steven R. Newcomb:
> Topic maps simply aren't
> meaningful unless each topic has exactly one changeless
> and eternal subject.

I think this might be a bit strict.  From a Platonist point of view all
abstract objects (Forms) are necessarily changeless and eternal.  However,
as humans do not have direct access to the Forms (shadows on the cave wall
and all that), getting the subjectIdentities right might be a bit tricky.
More seriously and practically, most of the things we want to talk about are
not changeless and eternal, at least from the human point of view (e.g.
people, countries, interchange syntax standards): to adopt Platonist
rhetoric, the abstract object/Form 'Tony Blair' may be changeless and
eternal, but the concrete object which is its manifestation is not.  In
these cases the above criterion would be more useful - while retaining the
intent - if it were weakened to something like 'each topic should have
well-defined borders of changeability' (if Tony has a liver transplant he's
still Tony, but if he has a brain transplant ... ?), but even this has the
potential to become a philosophical can of worms (Ship of Theseus and all
that).

Perhaps realistically the best we can say is that best practice is to have a
topic's subjectIdentity element point to some resource that everyone can
agree either is or refers to some entity.  The resource could be kept
up-to-date.  This resource could well be a topic element - in the absence of
direct access to Forms, what better way to define a subject (notion/idea)
than a comprehensive and authoritative listing of the subject's
characteristics.  The terms 'circularity' and 'infinite meta-level
regression' spring to mind.

No doubt such authoritative reference topics will emerge / are emerging.  In
the meantime - and even after - topics without subjectIdentity elements will
arise along with questions of if/how they should be merged.  These questions
are not technical.


Ivan Uemlianin, PhD
Head of Topic Map Development
Jura Technology Ltd

6 Tai Seion
Llanddeiniolen
Caernarfon
Gwynedd LL55 3AF


Head Office:
35 Norroy Road
London SW15 1PQ