Lists Home |
Date Index |
Simon St.Laurent wrote:
> email@example.com (Jonathan Borden) writes:
> >> Wow. This is where I really have to wonder if OWL is Karl Popper's
> >> revenge on the world, a naive view of useful information as a
> >> collection of facts, "objective knowledge".
> >Revenge? Is this a problem? :-)
> It is indeed a problem if you don't share the philosophy of knowledge
> underlying this technology.
We can debate philosophy of knowledge for eternity, the bottom line is that
there are software packages in existence that can accept certain tokens as
input and emit certain tokens as output, which people consider *useful*. In
particular the medical community has developed large classifications such as
SNOMED, Grail etc. which are based on this concept of classifications. Much
of day to day medical work -- particularly the back end coding and billing
departments -- are quite dependent on assigning codes to medical charts --
let's say this is a process of marking up documents. The tags used to mark
up are taken directly from hierarchical classifications.
> >Indeed it is common practice to informally refer to a particular
> >collection of triples as a "knowledge base"
> People even give presentations on "knowledge technology" and audiences
> listen. I don't know whether the problem lies with the speakers or the
> audience, but I try hard to avoid such conversation and consider it
> meaningless if not dangerous when I encounter it.
Of course this entirely depends on what you are trying to do when you sit
down at your desk everyday (or whereever and whenever). There are large and
well respected communities however, that don't consider this sort of work
meaningless, nor is there any real evidence that it is dangerous -- unless
of course if we consider that *misclassifying* something might indeed be
dangerous when the classification corresponds to a medical diagnosis.
> >I expect that any statement that is strictly limited to elements and
> >attributes has already been made, or else is a rehash of another
> >statement that has already been made, probably originating in the XML
> >1.0 rec itself modulo some issues with what really is UNICODE etc...
> >From a knowledge technologist's perspective, perhaps. I don't think
> conversation on using markup has even really gotten started, though, and
> there's certainly plenty of room for tools development that's barely
> been expored.
I'd label this as a converstation on "using markup", so hopefully its at
least as on topic as other converstations regarding abstractions such as the
PSVI none of which directly involve matched angle brackets or quotes.
> Given the choice of conversations, I'll stick to the strange little
> world of characters and markup rather than striving to build global
OWL is hardly about "global meaning", rather about quilting together small
patches of local meaning into useful constructs.