[sc34wg3] N0391-0394: New SAM/XTM documents

[Robert Barta]

On Sun, Apr 20, 2003 at 09:45:45AM +0200, Jan Algermissen wrote:

> Both, the SAM and the TMM exist independent from any serialized
> syntax (XTM, HyTM,..).
> 
> So, consider a topic map application/system that does not involve
> transforming serialized syntaxes, what is the TMM then?

It is a structure (read syntax) plus information what should happen to
ensure that a map is fully merged. So implicitely it defines the
operation

  fully-merged-map + topic -> fully-merged-map

It may have other constraints as well. You would call it "the essence
of what Topic Maps mean".

> And here is another question:
> 
> Is the entity relationship model a 'deserialization technique'?

If I interpret ERs as a set of schema definitions, then these have
no meaning at all. The classical interpretation of a schema set is
to have sets of tuples (one set for each schema). These are sets
and on the tuples generic operations have been defined:

   - projection
   - selection
   - natural, left, right joins
   - ...

Accordingly, this builds the semantics of one particular ER
definition.

Most modern interpretations of ERs are based on relational calculus
and not on sets of tuples. This means that every schema is interpreted
as a predicate over so many variables as the arity of the schema has.
The semantics of this is defined via the "models" (assignment of values
to variables).

Recent interpretations use ....tatata description logic, but this is
outside the scope of this discussion.

> And another one:
> 
> How do the entity relationhip model and the TMM differ conceptually?

Phew...

> [These questions might seem insane, but I have the feeling that your
> answers will help me to get the point]

... [stealing myself from this question ;-] "this should be clear from
the above", my professor would have said.

\rho