TM formal model (Re: [topicmapmail] multidirectional ...)

[Bernard Vatant]

BV
> | To go to the core of the debate, our graph searchers have always considered the topic
> | map model as unnecessarily convoluted and constrained in many
> | aspects. And basically I sort of agree with them :))

LMG
> In that case you should come out in the open with your arguments.
> We are defining the model now, and it would be better to hear your
> arguments before rather than after it is all cast in stone.

As a matter of fact, we are working on that right now. Without going to the details, the
model we will propose soon "cum commento" is a mathematical one based on hypergraphs. See
Patrice Ossona de Mendez and Pascal Auillans papers at XML Europe 2001 for mathematical
background.
This model is both more generic and formal than PMTM4 in the sense that it is completely
defined by set theory, and more constrained in a way. The added constraint, which the
current "implicit model" lacks, is the notion of distinct semantic levels, and the
correlative necessity to introduce constraints on stratification of the graph, using a
"semantic lift" to link levels together.
(e.g. to go from a topic to its role or type specification.) This constraint is the way to
escape the well-known "recursivity trap", and it allows to check for "well-formed" topic
maps (without semantic mismatch).

This model will seem at first sight far from the present ISO reflection track. We hope to
be able to explain it is not, and that it is really robust, and might even be a basis for
unification of TM with other "semantic" representations.

Bernard