Re: PMTM4 templates vs. TMCL (was: Re: [topicmaps-comment] RE: OASISvsW3C)

[Murray Altheim <murray.altheim@sun.com>]

"Thomas B. Passin" wrote:
[...]
> Yes, but this does not rule out having a schema model, for example, that
> extends what is already in topic maps.  The extensions become part of the
> dog food.

Yes, and unless someone beats me to the punch and creates something 
better than I have in mind, I'll be hopefully developing the ability to 
express these types of relationships using XTM syntax *because I need it*
and the CG community is currently not heading this direction. I'd started
on this last summer when working with the Cyc ontology but I need to do
a lot more homework before I tackle this one further. (The more you know,
the less you know, etc.)

> In CGs, a conceptual graph can only have conceptual relations that have
> their attached concepts.  But a conceptual relation is defined with slots or
> roles that are to be filled with concepts when instantiated.  To properly
> describe this, you need to introduce formal parameters (or lambda
> expressions).  I'd suggest that we will ultimately need formal parameters in
> topic maps, because otherwise we can't really define new associations
> properly.  The reason is that we want the definitions or templates to be the
> equivalent of classes rather than instantiated objects.  So we cannot refer
> to any specific instance in a template, and thus we need formal parameters
> to represent any and all instances.

I spent a couple of weeks working through the CG specification and related
documentation with the idea of developing an XML expression of CGs. Lack of
a formal data model and extant documentation on various components kept
this from occurring (though I did post a very rough first draft), and now
that John Sowa has pulled further development back into KIF and the ANSI
committee, it remains to be seen what will happen with CGs and XML. Peter
Becker, myself and a number of others are still interested in investigating
this further and I don't expect the idea to die.

> With formal parameters you ***would*** be able to use a topic map to define
> any new association or constraint you wanted to.  Without them, I'm not at
> all sure you really can.

I think that all you really need *to get started* is a set of PSIs for
a number of fundamental association (relation) types. I originally 
thought I'd pull these from the CycL ontology, which would be still
possible, but one of my slower-moving projects is to establish the set 
of these relations (ie., how much farther than first order predicate 
logic does one need to go? etc.) that would allow the core of either
a taxonomic or ontological system to be developed. This seems to 
immediately expand outward in different ways depending on domain, but
that core is still a goal, and I think it's doable. As was said recently
on the CG list, syntax is kinda irrelevant. All we need is 

    topic ~= concept
    association ~= conceptual relation

and we gain a bit of traction. Probably better than a brand new syntax
in some ways (we get to use topic map software to process it, and it
integrates nicely with the topic maps themselves).

just some ideas after a tiring day...

Murray

...........................................................................
Murray Altheim                         <mailto:murray.altheim&#x40;sun.com>
XML Technology Center
Sun Microsystems, Inc., MS MPK17-102, 1601 Willow Rd., Menlo Park, CA 94025

      In the evening
      The rice leaves in the garden
      Rustle in the autumn wind
      That blows through my reed hut.  -- Minamoto no Tsunenobu