[sc34wg3] SAM-issue term-scope-def

[Nikita Ogievetsky]

Bernard,

> * Nikita
>
> > Well ... computing is applied math.
>
> It should be at least ... sounds like the Universal Declaration of Human
Rights ... good
> to recall it from time to time, just to figure that nobody applies it in
real life :(
>
> > XTM is related to computing and some math here is not bad.
>
> Good to find support here from a true scientist :))

Whatever the word "true" mean :-)

> > I vehemently agree that topic maps have much less to do with logic and
> > inference than RDF, for example.
>
> Well, I tend to agree with Lars Marius that is mainly historical, because
> people who built RDF come basically from AI and logic community,
> so they care much more about it. But Topic
> Maps will need logic and inference at a point. That's what TMCL and TMQL
> are about, and we need that badly. If we say that TM have nothing to do
> with logic and inference, we shoot
> ourselves in the feet, and will never be heard from the above-quoted
> community. Any knowledge representation sytem needs some logical
> foundation at some point (hmm ... that
> sounds like John Sowa, I should stop here)

I always thought that Topic Maps provide a foundation
to which any type of logic can be applied.
Topic Maps (the foundation) in turn has its own foundation which we here
want to seek.
(better latter then never)

> > But there is certainly some set of axioms that we use to make our
judgments.
>
> Hear, hear. We need a topic map Euclides.

Exactly!

> > This set of axioms could be called "Processing Model".
>
> What has it to do with processing? Let's call things by their name. This
is a Formal Model
>
> > Interestingly enough it sounds that neither RM, nor SAM pretend to cover
it.
>
> SAM does not, it's clear. I thought that was the purpose of Reference
> Model, but maybe I was misled.

As I remember, RM provides a standard way for expanding Topic Map into a
graph.
All other models go on the top of it.

> > So may be there is a place for an independent "PM" model that should
> > answer questions like "what does <two lines are parallel> mean".
>
> Yes, if "meaning" means
> -- Formal definition of primitive objects
> -- Formal definition of relationships between those objects
> -- Formal definition of rules for those relationships

And :
-- rules of transformation,
-- congruency criteria
etc.

> ... what formal logic people call "semantics", I wonder why, because
formal definitions
> are useful just because there can't ever be any agreement on meaning :)
>
> > Actually at some point I offered a name "tom-tom" for it (theoretical
> > model  for topic maps)
>
> I buy this one! Is www.tomtom.org available?

Great! Unfortunately not (and it was not a year ago:-))

> > also because construction of some tom-toms resemble graphs - a set of
> > connected drums.
>
> Basically, rarely more than half a dozen of them. Maybe we don't need
larger TM after all
> :)
>
> > I think Bernard's questions should be a good start.
>
> That's basically what I am good at - asking relevant questions, hoping
> others will answer them.

And then trying to answer them yourself? :-)

> > TMPM4 was about it, but RM jumped on a lower level (or I am missing
> > something).
>
> What do you mean a "lower level"? It figures nobody is much eager to
support RM any more
> ... Are Steve and Michel still around?

Oops! I did not say that I do not like RM. I said that it does not pretend
to provide
complete set of axioms for Topic Map processing.

--Nikita.