RE: [xtm-wg] parallel development of syntax and concept models

["Michel Biezunski" <mb@infoloom.com>]

Steve has provided an excellent contribution to the understanding of
what's going on with the model we use for topic maps. Although I agree
with him on what he is saying, I have a different way to express it.

One model, Two Perspectives
========================

Instead of speaking of the "conceptual model" versus "syntax", I would
rather insist on the fact that there is only one model, and two perspectives
: the
"Foundation perspective" versus the "Hyperlink perspective".

The Hyperlink Perspective
=====================

The "Hyperlink perspective", also (abusively) called "syntax", says
basically: Topic maps are made of 2 hyperlinks: one is called topic,
another one is called an association. The topic hyperlink has a
supplementary property: it has name(s). The association hyperlink has
a supplementary constraint: it can only connect topic links. Both
hyperlinks can be scoped. That's basically it.

[I have excluded facet, which is a 3rd hyperlink, because facets don't
belong directly to the topic map architecture. ]

The Foundation Perspective
=======================

The "Foundation perspective", also (abusively) called "conceptual
model", considers instead of hyperlinks the individual objects
connected together as "binding points", the connector linking them to
other binding point being considered as a property of each of these
binding points. It's the exploded view, as opposed to the factorized
view (in the hyperlink model). This view is consistent with the grove
paradigm
and can be described using property sets.

Advantages of each perspective
===========================

What is essential is to understand that there is only one model for
topic maps, and two expressions of it. The two expressions are
functionally equivalent.

The Hyperlink perspective has an enormous advantage: it's extremely
simple, compact, and powerful. With two hyperlinks, it's possible to
capture very complex knowledge representations. Because of its
semantic neutrality, it applies to an amazingly great number of cases
in real world applications. The hyperlink representation (based either
on HyTime or XML-link) is the building block necessary to represent
what's needed here.

The Foundation perspective came from an attempt to represent the topic
map information once exploded in the form of a grove. It expresses the
basic properties attached to every object resulting from this
explosion. The Foundation perspective can be seen as a checking
mechanism. By expressing the topic map model this way, it helps
ensuring that all details are correctly expressed. One example is
the fact that it seems that display name can be used as well as base
name to merge topic maps. It is (as I recall) the intent of the
original authors of the standard to use base name for merging rather
than display name (which should be used exclusively for display).

It is precisely for the reason that the work on the model is helping
showing what interoperability problems there might be if we are not
precise enough that this work is essential.

Topic Maps as a Guinea Pig ?
========================

I think that using property sets and groves to arrive to an exploded view of
a
model is something that applies not only to topic maps but to a wide
variety of models and languages, including XML or SGML
themselves. It happened that topic maps have been chosen to experiment
on this idea and that this is benefitial, but only until a certain point.

We should refrain from considering that topic maps is the only (or
privileged) guinea pig
for groves and property sets. These world views are of general applicability
and
they are slowly being recognized as a valid tool to see the world, provided
they
are considered to help and not to prevent things to happen. I urge those of
us who
are interested in promoting these ideas to try to use them pervasively in as
many
application domains as possible.

Arguments for adoption
===================

I think we should learn the lesson of this current discussion. It looks like
the relationship
between the two perspectives is far from trivial. If we require from topic
map adopters that they understand all the nuts and bolts of this double
perspective prior to adopt the
spec, then we're heading for trouble. People will start saying: this thing
is too difficult to grasp,
let's look for something simpler, which is exactly what we want to avoid.

If we don't arrive to a common clear understanding of what we do and why and
how
we do it, and what are the frontiers of our objectives, then we'll have
exactly the same
problem as RDF people have, i.e., they are trying to define exactly
the same problem as we do, but their perspective is too close to the
exploded model, and therefore, users are not seeing the applicability
issues. It's like building a house and considering only the materials,
such as stones, pieces of wood, bricks, without considering that we
are building a living room, a kitchen or a bathroom. A house must be
correctly built, but to be salable, each room must be recognizable for
its function. That's the challenge, and we all need to be aware of it.

Michel

==========================================
Michel Biezunski, InfoLoom, Inc.
Tel +33 1 44 59 84 29 Cell +33 6 03 99 25 29
Email: mb@infoloom.com  Web: www.infoloom.com
==========================================


-------------------------- eGroups Sponsor -------------------------~-~>
GET A NEXTCARD VISA, in 30 seconds!  Get rates
of 2.9% Intro or 9.9% Ongoing APR* and no annual fee!
Apply NOW!
http://click.egroups.com/1/7872/4/_/337252/_/967440659/
---------------------------------------------------------------------_->

To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com