RE: SUO: RE: RE: The Story So Far
At 07-03-01 16:12 +0100, West, Matthew MR SSI-GREA-UK wrote:
[...]
>So let us take the case of a waste bin and a desk. I would expect that a
>good classification system would enable me to store and retrieve the
>information about these two objects so that I could answer questions like:
>
>"Can I put the waste bin on the desk?"
>"Can I put the waste bin under the desk?"
>
>(depends a lot on the size and weight of the waste bin)
>
>On the other hand I think the idea here is to try to get to the stage where
>you can simply ask the questions directly to the "system". This means that
>you need to have captured a lot more in terms of general knowledge about how
>the world is and how it behaves - so called "common sense".
>
>So this would mean that whilst classification systems were certainly part of
>what is needed, there is some other stuff as well, which is normally just
>left in the definitions of the classes, and often not even found there, that
>needs to be formalised and made explicit.
[...]
### Matthew,
### A very clear statement and example! For me OK if I can assume that the
operative word in both questions above is "Can". I would like to see
ontologies -in practice, that is- as (mathematical) objects that describe
"allowed" or "plausible" situations/relationships/... I hesitate to use the
term "possible worlds" since that has become laden with different meanings,
although e.g. Kripke semantics (or even better perhaps, Dunn semantics as
John Sowa recently has pointed out to me --see his website) do seem rather
nicely to suit this idea, one of the things we are currently investigating
here in our DOGMA Project. (Watch this space!)
### To elaborate on the "in practice" bit: the fact that a waste bin is_a
bin is_a receptacle is_a ... is_a physical object, for instance, and
analogously for a desk, and that one physical_object MAY be
under/on/next/... another one, seems the right ontology stuff to me. An
application could consult this o'gy (3D, 4D, ...) e.g. to check that it has
not introduced contradictory spatial relationships. But you correctly say
that weight, size, etc of these objects play a role here, they CONSTRAIN
the allowed relationships/situations/... and as such it is my belief such
rules belong in another "layer" of o'gical knowledge, and certainly not in
an "upper o'gy", perhaps even in some mediating object between the o'gy
proper and the application. This is good database practice by the way, as
you undoubtedly know.
### I realize this mainly is a methodological issue, but that is my
hobby-horse in this affair... :-)
--Robert Meersman
----------
Prof Dr Robert A Meersman VUB (Vrije Universiteit Brussel)
Department of Computer Science STARlab --Building F-G/10
Pleinlaan 2 B-1050 Brussels Belgium
phn (+32)(0)2 629 3308 fax (+32)(0)2 629 3525