[topicmapmail] Conceptual Graphs are Step 6

[Dan Corwin]

Jack Park wrote:

> Useful thread, this.

Thanks, Jack.  If you join in, I'm certain it will be :-}

> I suspect that the new RM is much closer to the right modeling tool for 
> doing CGs and having topic maps fall out simultaneously, than is XTM.

I would agree in principle.

But precisely because it is "new", I think RM is still a bit mysterious
to many people.  So to many readers of this list, talking about CGs
using the RM would probably hurt overall clarity, not help it, or at
least split their attention up between two competing subjects.

Similar problems already cripple most discussion of CGs because their
details soon turn to FOL - a language family as foreign to most people
as those in Africa.  CGs are confusing *mostly* due to this FOL-speak.

If you drop the FOL from CGs, you get a graph language based on the
types of associations and roles used in English sentences - a very
useful tool for building highly intuitive semantic nets, which with
just a few standards added could become very efffective topic maps.

The specific role types and their meanings are the core of what CGs
have to say, and they would become the heart of any XTM-CG standard.
A good introduction to such types, with minimal FOL-speak, is:

   http://users.bestweb.net/%7Esowa/ontology/thematic.htm

XTM 1.0 and a few PSIs are *all one requires* to express such graphs
formally and portable as associations any topic map engine can load.

But for readability and brevity on this thread, I'd hope most CG models
we discuss here might actually be presented in LTM, not XTM.

Either notation has the big advantage that anyone on this list can
follow it and contribute.

But I hope you will keep us honest, Jack.  If anything said later
needs to be restated in RM terms, please do help us out on that.

Cheers,
Dan Corwin