[tmql-wg] Every thing is a 'thing'

[Robert Barta]

On Fri, Mar 09, 2007 at 01:28:58AM +0100, Lars Marius Garshol wrote:
> >What about something more philosophic ;-)
> 
> I've no trouble with that, but a new subject would be nice...

Sorry, will do.

> >   Now, what should a TMQL processor deliver here:
> >
> >      select $thing
> >      where
> >         $thing isa tm:subject
> 
> There are several possibilities here:
> 
>  (1) All topics in the topic map
>  (2) All Topic Maps constructs in the topic map
>  (3) All constructs + all values in the topic map
>  (4) All constructs + all values in all value spaces for all datatypes
>  (5) All values that could theoretically exist
>  (6) An error :-)
> 
> Strictly speaking (3) is the most logically consistent. What we  
> actually want is (1), though.

Right. Or we could take the viewpoint that

  where
      $thing isa tm:topic

gives all topics and

  where
      $thing isa tm:subject

gives (2) or (3) above. Question is whether values are 'subjects' in
the TMDM sense.

> >   Structured Discussion
> >   ---------------------
> >
> >      ? Should all map items be an instance of tm:subject implicitly
> >        + tm:subject is a great placeholder
> >        - TMDM does not say it (or does it?)
> 
> Whooops. Does this mean that you intend for it to be (2)?

I'm just asking. If atomic values _are_ subjects, then it would be
(2), right? And if not, then probably (3), I guess.

> >      ? Should all topics be a subclass of tm:subject implicitly
> >        + tm:subject is a great placeholder
> >        - TMDM does not say it (or does it?)
> 
> Here it's (1), right?

The question is what should

  (a)    select $x    where       $x isa tm:subject

  (b)    select $x    where       $x iko tm:subject

return.

> And, no, TMDM does *not* say this. The prose implies it, but there is  
> *no* formal machinery that does anything like this.

I think we shortly touched this one in Leipzig, but I am unsure
whether this was followed up by anyone:

  - is an association a specialization of subject?
  - is a topic a specialization of subject?
  - is an occurrence (and a name) a specialization of association?

Things like these. So something like an "Ur-Ontology" for TMDM. I
faintly remember that you had a blog entry in this direction but
cannot reproduce it now.

But once this is written somewhere (TMDM, TMRM, TMQL), then we would
have a commitment.

\rho