[sc34wg3] Topic Maps land and SAM land
Lars Marius Garshol
10 Feb 2003 17:53:01 +0100
* Michel Biezunski
| RDF proceeds with statements. It is directed. There is an origin and
| an end. It's difficult in RDF to say something about the same given
| subject involved in another statement because there is no built-in
| solid mechanism to make these subjects the same.
Uh? If the URIs are the same, RDF considers that the statements are
referring to the same resource (subject). There's also machinery in
both DAML+OIL and OWL for working out when different URIs represent
the same thing, and even for working out when they must represent
different things. So the situation compared to topic maps is that you
have a stronger apparatus for declaring when things are the same, and
in addition an apparatus for declaring when they can't be.
The similarity with between the RM and RDF goes as follows:
- subject (in RM) the same as resource in RDF,
- proxy (or node, or whatever RM calls them right now) is the same as
RDF node in RDF,
- assertion in RM is the same as statement in RDF.
There are some minor differences with regard to how data values get in
there (node properties in RM, literals in RDF) and a not-quite-so-minor
issue with respect to reification, but that's really it. In all, the
similarities are much greater than the differences, which is not the
case if you compare RDF/SAM. RDF is lower-level than SAM, but pretty
much at the same level as RM.
| RDF is built on a predicative logic in other terms.
Not really. It's a graph. The logic layer is built on top of the
graph. If you look at the RDF Semantics document you'll see that what
it describes is a logic built on a graph syntax.
| The fact that RDF is a graph and TM is expressed as a graph is a
| side issue.
| It's my impression that you are putting more focus on formalism than
| on what is expressed.
Not correct. I would have said the same thing had the RM or RDF been
specified in terms of the infoset model, as in fact they can both very
easily be. What I said was based on my analysis of RDF that I
presented in Orlando, and which was the basis for the RDF->TM mapping
that underlies Ontopia's autogen toolkit.
| This is why I am advocating for getting to the proper level of
| abstraction. Let's forget the formalism for a moment, and discuss
| the content.
I've been doing that all along.
| If you do that, you'll see that RM *is* TM -- not in the sense that
| it is "on the TM side" but it is what TM is. Same thing with SAM.
| SAM *is* TM and actually reflects faithfully a number of
| implementations of it.
Yeah, I know you think this, but I don't.
Lars Marius Garshol, Ontopian <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50 <URL: http://www.garshol.priv.no >