ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Looking forward at the past

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Fri, 10 Oct 2008 19:59:50 EST
Message-id: <48efec76.11a6d.0@xxxxxxxxxxx>
Pat and Don,    (01)

I partially agree with that point:    (02)

PC> When you try to "interoperate" with a plumber or any other
> person, your chances of achieving successful communication
> depend on both of you already sharing a very large basic
> vocabulary.    (03)

But I'd replace "very large" with "an amount varying from
almost nothing to a great deal".  It would be more accurate to
say "common assumptions about the subject domain", independent
of whether those assumptions are stated in a common language
or even a language of any kind.    (04)

I also agree with the following point, but first we have to
decide how to organize that foundational ontology.    (05)

DC> The reason you can effectively communicate with the cast of
> characters in your examples requires much, much more than the
> lowest level task common definitions. Many of these things are
> of the sort that would be in a foundational ontology.    (06)

As I have said in various notes, my view of a foundational
ontology consists of a large, but very sparsely axiomatized
type hierarchy, which would primarily be based on two relations:
type/subtype and part/whole.  In fact, it might be better to
think of it as little more than a well-designed taxonomy and
meronomy for a large vocabulary.    (07)

In addition to those very sparsely axiomatized terms, there must
be specialized chunks or modules or microtheories of knowledge
with much more detailed axioms.  I think Don's list is fine,
and it could be enlarged with a long list of esoteric areas that
may be critical for certain applications.  But in many cases,
one party to the dialog may have vastly more knowledge about
these esoteric topics than the other.    (08)

DC> * Time
    * Location
    * Interdependencies in general and specifically for each domain
    * A host of concepts:
       -contracts
       -Residential building codes
       -budgeting
       -competitive pricing
       -anatomy
       -nutrition
       -locale specific cuisine
       -context to parse the communications correctly    (09)

But note that these come into play in *different* tasks with different
people, who may have different kinds and amounts of knowledge about
the shared task.    (010)

A contractor usually has far more knowledge of building codes than
the house owner and a dentist has vastly more knowledge about dental
anatomy.  In many cases, interoperability is successful *because*
the amount and kind of knowledge is *asymmetric*.    (011)

To return to my original point:  When talking with the plumber,
dentist, waiter, clerk, surgeon, contractor, etc., the amount of
shared vocabulary that the two parties require is generally a
tiny fraction of what either one of them knows.    (012)

Furthermore, the core intersection of vocabulary required for all
six pairs of interactions is very tiny.  The core intersection
is seldom, if ever, verbalized.  It is the common knowledge that
dictionaries do not state explicitly.    (013)

These are the reasons why I recommend a very sparsely axiomatized
taxonomy and meronomy of terms.  (Note that the definitions in
Longman's dictionary leave out an enormous amount of information
that would be required to use those terms effectively.)  The basis
for detailed reasoning is in the very specialized microtheories,
of which one party may often know a great deal more than the other.    (014)

Essential point:  Many, if not most, kinds of interoperability
are asymmetric:  one party may have vastly more knowledge about
the domain than the other.  Even when the amounts of relevant
knowledge are similar, the two parties may have very different
kinds (e.g. manager and employee, contractor and subcontractor,
CEO and accountant), and they communicate primarily on their
common intersection.    (015)

John    (016)

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (017)

<Prev in Thread] Current Thread [Next in Thread>