[topicmapmail] Subject Identifiers metadata

[Dan Corwin]

*** Kal Ahmed:

> I was talking about the thing that we conceptually use to represent a
> subject - irrespective of its representation in some computer system. We
> conceptually require this thing as a place to hang our statements about
> the subject (assertions), but in itself it provide *no information at
> all*.
> 
> Anyway, all this was only to say that I don't believe there is a subject
> "the topic foo" that can be reified - you can of course reify the
> <topic> foo or the topic record foo in a database and it is for these
> concrete representations that we may want to provide metadata.

If there truly were no such subject, Kal, then how can you discuss it
so easily with us in both of your paragraphs above?

We all need the ability to do this in TMs as well as email, both for
the abstract type and any of its instances.


*** Thomas B. Passin:

 > .. if you think of a particular subject identifier as a notion in
 > its own light, then you can create a topic to represent (the idea of)
 > that identifier....

Right, Tom, and assign that abstract type a highly visible PSI, saying
that if anyone needs to reify any instance of such a type ...

 > ... The type of the [reified] topic specifies that it is intended to
 > represent a subject identifier, the value of the subject identifier
 > could become the subject indicator or be captured with an occurrence

After re-reading this plan, I finally got it.  Sounds just right to
me, and elegantly answers Kal's original question on this thread.

To reify any subject indicator, you just *declare* its new topic to
have the "subjectIndicator" type, and nobody gets confused.  Bravo!

It also easily generalizes to allow *topic reification* in the same
easy way: just declare the *type* of the new topic to be "Topic", and
use the correct PSI, and there cannot be much confusion about what it
represents, even in XTM 1.0 file format.

Which means it takes no new XTM syntax to implement this plan, just a
couple of new or modified PSIs citing "best practice" documentation
about the need to declare types carefully when doing reification.

Internally, would either of your TM engines need to do anything at all
in a different way if this plan were adopted?

Dan Corwin