[sc34wg3] TR: comment - RDFTM: Survey of Interoperability Proposals

Murray Altheim sc34wg3@isotopicmaps.org
Thu, 10 Mar 2005 10:23:36 +0000

Robert Barta wrote:
> On Thu, Mar 10, 2005 at 02:31:06AM +0000, Murray Altheim wrote:
>>So I think we can relax in knowing that *at least* as much
>>thought has gone into modeling Topic Maps as was RDF, and we've
>>spent a great deal of time on issues like identity from a real
>>user-centered perspective (not particularly from logic, which
>>is not even demonstrably useful in a world that doesn't always
>>follow logical rules, i.e., the world we live in). Perhaps one
>>day we'll be so likewise fortunate to have a logician grace us
>>with a model so that we can strut our stuff too.
> [ long and interesting post snipped ]
> Murray,
> I would agree that asking for a formalized model just for the sake of
> formalization is an rather empty goal. Many people have, justifiably
> dismissed it as an academic execise, and I would normally agree,
> unless...
> But then we realized that a sound mathematical model [1] allows you to
> define operations on topic maps and build - based on these operations
> - query and constraint languages.
> For TMQL, for example, the semantics is not only described in prose
> (argh :-), but also - via several layers - in terms of that formal
> model.
> If that is not practical enough, the query processor which I prototype
> on the weekends (hey, this is what weekends are good for, come on) is
> using the formalism to make transformations of query statements
> maintaining semantic equivalence. Whole optimization techniques can be
> described by that. Given that TM processors are prone to be slow in
> operations this may help.
> And having the potential to map query/constraint expressions
> automatically into FOL expressions is cool. Fullstop. :-)
> \rho
> [1] http://astma.it.bond.edu.au/junk/tau-model.pdf

Hi Robert,

I agree entirely with what you're saying, the only issue is that
one must then assume that the people supplying you with the Topic
Map documents also understand and are following the same model in
their designs, which is a much bigger issue, one of agreement and
intercommunication. Availability of a model is pointless if those
creating document types or documents (i.e., applying the modeling)
either don't understand or don't follow it. So for anyone who's
read Pat's RDF Semantics document and claimed to understand it,
then hey, by all means, share away. But I kinda doubt those
developing most RDF applications have or do, especially as the
document is stated in a mathematical language most RDF developers
don't understand, and is also rather recent. RDF Schema was
written long before Pat published his first draft, so I'm curious
as to how it fits. It probably doesn't. RSS? XUL? Hah! I even
wonder how OWL fits into it. When you retrofit, sometimes you end
up with those little loose screws on the counter, and then wonder
where they came from. I can just say I'm glad I'm not the one
doing *that* analysis. What a headache.

I think anyone who claims that those passing around RDF content
are basing their decisions on the presence of some formal model
are making similar claims, and falling into a similar trap. There
may be *many* such models for RDF and Topic Maps, and indeed, my
point is that if we're going to be mapping Topic Map documents to
FOL, we should probably be mapping them via Common Logic, which is
being written by people like Pat Hayes, John Sowa and Chris Welty,
all some of our more public experts in applications of FOL. Pat
has gone to the trouble of learning XML markup (much against his
will and better judgment) in order to help out the markup community,
and I aim to do whatever possible to expedite its availability and
use within the Topic Map community. I've been advocating an "XCL"
for the past few years (and an "XCG" for longer), and it will now
come to pass. It has the advantage that if someone as clueless as
I am can understand it, it is likely to be taken up more widely.

I wish I had the time to build a CL-based inferencing engine that
could import XTM, but I'm sure somebody will do it once the mapping
to CL is available. Certain parts of CL are already starting to
show up in Ceryle, though hidden behind the scenes. At that point,
all of this stuff just becomes an alternate syntax to anyone
following the model (though not to just anyone using the syntax,
a common fallacy), and all the tiresome religiosity can (hopefully)
go away.

Given that we'll all also be able to point to an ISO standard for
that expression of FOL, we'll all benefit.


Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

  "However, Mr Moeliker was informed of an American case involving
   a squirrel and a dead partner, although in this case it is not
   known whether the necrophilia observed was homosexual or not as
   the victim had been run over by a truck shortly before the
   incident." -- Necrophilia among ducks ruffles research feathers