[topicmapmail] DRAFT: Datatypes for XML Topic Maps

[Murray Altheim]

Kal Ahmed wrote:

> Murray,
> 
> This is interesting work. There are a number of things which I would like to 
> hear your thoughts on.
> 
> Firstly, have you looked at any of the proposals for structuring scope [1] 
> [2]. I think that a structured scope is needed for your proposed datatyping 
> mechanism to work. 


Not until you pointed to them -- thanks.

My feeling about these types of interpretations is that the kinds of
problems delineated in Marc's paper will continue until a more proper
solution is found, but not one in syntax. What I mean by this is made
clear if one is familiar with Conceptual Graphs. If we were to solve
one or two problems by coming up with a new syntax or make a few rules
for interpretation, we'd only solve a very small part of a larger
problem.

What we really need to solve this is yet another thing I began working

on over a year ago, in those early, aborted efforts to implement PSIs

for Cyc's "Predicate Logic" or represent CG in XTM, i.e, we're beginning
to travel in the realm of things like KIF, prolog, etc.

We need the full onto-ourage of first order operators, not just union
and intersection, but things like "and", "or", "not", "implies" (material
implication), "there exists" (existential quantifier), "for all" (universal
quantifier), "equals", "equivalent", "different", "true", "false", etc.
The last two are at least provided by the datatypes PSIs, but we're
fooling ourselves trying to represent common logic without something
like Common Logic. Honestly, I'm waiting for Sowa et al to publish a
spec for CL, and one of the first things I plan to do is create a set
of PSIs from that.


> Secondly, have you given any thought to ways in which derived simple datatypes 
> and complex could be built from the datatypes you have presented here. 
> Intuition tells me that a "derivation" and "restriction" association type 
> could assist in creating derived simple types, and that reification would 
> play a role in building complex datatypes...


Certainly, but as I hope I've made clear above, absent a more complete
set of PSIs for logic, I think it'd only be a stop-gap solution.

But I am interested in the idea of representing complex types either
in XTM syntax (verbose and without logic PSIs, prone to interpretive
errors, I *think*), or perhaps a CL+XTM, or perhaps a mix of XSD
datatype syntax for complex types mixed with XTM in some way. There's
also the idea I mention in the Datatypes spec about stealing more of
the UN codes, this time for measurement. It'd be nice to have the
metric system of units as PSIs, for example.

I think we may all (myself included) be rushing into a well-explored
territory with a new tool set under our belt, perhaps ignoring many
who've gone before. Given that this is a criticism I've leveled against
the W3C on occasion, I'm trying to avoid doing that myself, instead
looking carefully at existing work as much as possible.


> Cheers,
> 
> Kal
> 
> [1] Structuring Scope, Marc de Graauw - 
> http://www.marcdegraauw.com/files/structuring_scope.htm
> 
> [2] A Proposal For Extending Scope In XTM, Kal Ahmed - 
> http://www.techquila.com/scope-proposal.html

Murray

......................................................................
Murray Altheim                  <http://kmi.open.ac.uk/people/murray/>
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK

            If you're the first person in a new territory,
            you're likely to get shot at.
                                                     -- ma