Pat, (01)
I agree with your litany of challenges facing Cross Domain
Semantic Interoperability (CDSI). (02)
My position is not that we 'can' solve this, but that 'maybe' we
can solve it (in 5, 10, 20??? years), and since it is such a big problem
and bigger opportunity, wealthy organizations should seriously try to
solve it. (03)
Jim Schoening (04)
On Wed, 22 Nov 2006 14:40:43 -0600 Pat Hayes <phayes@xxxxxxx> writes:
.
>
> The problem with things like these philosophical
> kind of divergences at the upper-middle level is
> that ordinary ideas like 'physical object' or
> 'human' turn out to have different properties in
> the different frameworks. For example, a
> DOLCE-style ontology ( or one based on Barry's
> snap/span contrast) will distinguish 'human'
> meaning a continuant (I am the same person I was
> ten years ago) from 'human' meaning a human
> lifespan, whereas a differently framed ontology
> would not make this distinction. So, is the set
> {Pat, Pat's life} a singleton or not? I suggest
> the right answer to questions like this is, you
> can have it both ways. If anyone wants to make a
> distinction like this, they are free to do so. If
> someone else wants to not make the distinction,
> then evidently what they mean by 'human' isn't
> the same as either of these, so now we have a
> three-way distinction: human-A-continuant,
> human-A-occurrent, and human-B which isn't either
> an occurrent or a continuant. (Maybe its a kind
> of blend or sum of these ideas, but that way of
> putting it isn't very helpful as A thinks this is
> logically incoherent and B doesn't know what
> 'continuant' means in the first place. B might be
> thinking about 4-d worms, or some such idea: the
> A notion of continuant doesn't make sense in that
> way of thinking.)
>
> Each of these inherits using the higher axioms
> from its own ontology, but you had better not let
> them cross-inherit, or everything goes
> pear-shaped very quickly. So then we need the
> bridging axioms to connect these together
> appropriately: but until you agree to accept the
> distinctions, you can't even get the axioms
> written: if you try to get rid of the
> distinctions by generalizing upwards , or by
> asking everyone to please stop arguing and agree,
> you will never get the problem solved. Its
> important to let 10|3 flowers bloom, then find
> ways to cross-breed them.
>
> But note, its not that these A and B folk differ
> about anything that a philosophical layman would
> consider to be a question of fact about what
> 'human' really, in some sense, means. They all
> would say, when faced with an actual human, that
> this is what they are talking about. They all
> would say that they are just writing down
> ordinary everyday common sense knowledge, and
> they would all be right. Its just that they would
> encode this same ordinary knowledge into
> different logical/philosophical/ontological
> styles when they go to the formalism.
>
> >I am not worried about global consistency
>
> :-) I am. I have no idea how to rationally handle
> global inconsistency, and I havnt heard a
> sensible proposal for how to do it. And its very
> easy to arrange global consistency (assuming
> local consistency) so why not just assume it and
> move on?
>
> >and in fact can tolerate two
> >domain ontologies that use the same middle/upper to be mutually
> >inconsistent.
>
> Its easy to make them consistent by using
> disjoint vocabularies. If there are three (or
> seventeen) notions of 'human', use three (or
> seventeen) different logical names to denote
> them. Or use a more systematic way of assigning
> meanings to names, e.g. by using contexts or
> contextualized names.
>
> >But they will only be able to share the consistent
> >information (or the one's inconsistent info will have to be
> treated
> >somehow special when brought into the other).
>
> Well, that 'treating special' is what Im
> suggesting to do as a general policy, from the
> get-go. Rather than think of merging ontologies
> as a kind of addition, and an inconsistency as
> something that requires triage or repair, think
> instead that one would not expect that ANY
> concepts from one coherent ontology will be
> exactly identifiable with ANY concept in another.
> Treat them as strangers, then introduce them
> carefully. If they do turn out to be the same,
> its very easy to write the bringing axiom:
>
> (= TomatoInA TomatoInB)
>
> Pat
>
>
_________________________________________________________________
Message Archives: http://colab.cim3.net/forum/cuo-wg/
Subscribe/Unsubscribe/Config: http://colab.cim3.net/mailman/listinfo/cuo-wg/
To Post: mailto:cuo-wg@xxxxxxxxxxxxxx
Community Portal: http://colab.cim3.net/
Shared Files: http://colab.cim3.net/file/work/SICoP/cuo-wg/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/CommonUpperOntologyWG (05)
|