[sc34wg3] Topic Maps land and SAM land

Lars Marius Garshol sc34wg3@isotopicmaps.org
10 Feb 2003 09:02:28 +0100

* Lars Marius Garshol
| The two models are different. Either you implement the SAM model, or
| you implement the RM model.

* Sam Hunting
| There's that "either/or" thing again. 

Look, for me to implement the RM model means to create something that
actually maintains a graph consisting of nodes and assertions
structured according to sections 3 and 4 of the RM4TM. If you do that,
you have not based your implementation on the SAM, but are using a
different model. Similarly, if you create an implementation whose
internal structure consists of topics, names, occurrences,
associations, and suchlike, you are implementing the SAM model, but
not the RM model. 

You can't do both at the same time, although you can certainly have an
implementation that is based on one of the models, and then use that
as a basis for implementing the other. That implementation, however,
is still very much based on only one of the models, and the
implementation of the other model is very much secondary.

Let me put it this way: when Tony Coates says he has created an RDBMS
schema for the RM rather than for the SAM, why don't you tell him that
he is mistaken? The same applies to yourself when you say that the
GooseWorks toolkit is based on the RM.

Don't get me wrong: I think Tony Coates is right, and I think you are
right about the GWTK. But obviously neither implements the SAM model
directly; they implement the RM model. Similarly, the OKS, TM4J, and
K42 all implement the SAM model (or, rather, something very close to
it), but not the RM model.

How it's possible for this to be controversial I am not sure. Maybe it
has to do with the confusion around what it means for the Reference
Model to actually embrace everything.
* Lars Marius Garshol
| If the OKS is an RM tool, then Excel, Notepad, and Apache are also
| RM tools.
* Sam Hunting
| Well, supposing that all of these tools generated data that
| contained inherent topic map information, and supposing that they
| could be specified as such using the RM, would that be such a bad
| thing? 

The question is whether they actually use the RM model. I don't think
any of them do. None of them follow the RM graph structure. Therefore
they are not RM tools.

| "Embrace and extend" sounds like a pretty good idea to me ;-)

It sounds like empty words to me in this context.

Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >