[topicmapmail] Anyone looked seriously at XFML?

[Murray Altheim]

Thomas B. Passin wrote:

> [Murray Altheim]
> 
> 
>>I'm specifically curious if XFML really provides anything that
>>XTM couldn't with some PSIs provide. I've been looking into
>>faceted classification schemes for the last year or so and would
>>rather use an existing technology within my application than use
>>something both new and different, if possible. If I don't use
>>XFML I'll likely develop a faceted classification/grounded theory
>>schema of PSIs for use in my existing XTM/visualization code.
> 
> Interesting.  My take on it is this. In the same way that Topic Maps is a
> sort of design pattern for knowledge representation - one pattern out of an
> large and indefinite number of possible patterns - that xfml is likewise a
> design pattern.  As their web site points out, you could regard it a a
> subset of topic maps.


Yes, I agree. There are many ways one can look at topic maps, particularly
XTM, eg., as a graph structure to "overload" with whatever semantic
structures as necessary, as a mapping formalism, as a syntactical basis
for ontological structures, any mix of the above, etc. etc.  XFML is a
very specific markup language incorporating ideas from faceted classification.
How faithful they've been to FC I'm not certain, nor am I on whether a simple
implementation of FC semantics in XTM would perhaps suffice better for my
purposes (which admittedly aren't the same as everyone else's).


> I noticed one thing xfml has that Topic Maps do not.  They have an
> occurrence strength. I have often though that it would be good to indicate
> the strength of the players in an association, because I think that in real
> life the strength of a relationship plays an important part in our thinking.
> In fact, I am sure that we can temporarily change thoses strengths to let us
> focus on different aspects of a situation.


The approach to "strength" of course varies enormously upon domain -- it
might be called "confidence level" or other terms. As you might be aware,
I've recently posted a prototype topic map for datatyping, so that one
could (for a given domain) create a metric by merging say, "Integer" or
"Floating Point" with "Confidence Level" to characterize a value. With
XSD's ability to add facets for minimum and maximum values, this could
be quite functional, again, within a specific domain. I don't think this
is really generalizable.


> I think that the xfml restriction that a given topic can only be in one
> hierarchy would seriously hurt most of the uses that I am interested in, but
> for some other uses it might be quite good.


Yes, that's one aspect that really kills it for me. I'm not sure
if this is due to their using topics as facets (which must in FC
be exclusive or the scheme won't work), or some other reason. I'd
get on their mailing list but I refuse to get near Yahoo Groups.
Too bad.

Murray

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

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