Mike --
As I mentioned, we are dealing with very practical,
not philosophical issues. There are two kinds of
things computers can do for us: interpret and act on
communications, and perform robotic activities.
In dealing with the interpretation of communications or
input data that a person created, a stationary computer
will only need to be able to determine what was in the
mind of the creator of the document or data in order to
respond appropriately. That is the issue I discussed in my
reply. These representations will generally include the context
and of course they commonly refer to physical situations in the
world, and will include things that can only be partially defined.
But the only physical contacts that our desk computers have is
with the peripherals from which they will get information and
to which they will transmit information. The conceptual structures
in the ontology will include information on how interactions
with those peripherals relate to the conceptual structures
(e.g., get an invoice, find the amount due, print a check.)
Other than the peripherals, the computers we are concerned
with at present do not interact directly with the external
world, only with representations of the external world
that people create based on the representations in
our minds. In this case the "true meaning" is indeed
what was intended by the person creating the communication.
Certainly, deciding how best to represent physical entities
in abstract conceptual structures is a very relevant
issue, even if if were possible to divorce it from simply
trying to reproduce the way people have already represented those
physical things in their minds. But after many years of debate
it should be quite clear that the only objective way to know what
manner of encoding is better than another is to see how applications
using that encoding perform on real tasks. It won't be
resolved by debating at a workshop. It's been tried.
Only when we get to robotics or interpretation of
physical sensor information will it be necessary for the
computer to form its own model of the external physical
world, rather than depending on the model encoded in its
own ontology and in the minds of those agents with which
it is in communication. Robotic agents with those kinds of
sensory/motion capabilities will then need some
ability to form models of the external world based on
input from its sensors, and it will be in a position to
experimentally verify or falsify theories and models of
the external world. Then you get into the question of how
to determine whether the conceptual structures faithfully represent
the physical objects and processes. This is an interesting topic,
but it is not necessary to address those issues when
dealing with information-processing tasks such as
the interpretation of communications. There are research
projects in which the creation and verification of models
of physical situations are investigated in the context
of robot action. We aren't dealing with those at present.
For this group at this point we do not
anticipate that our computers will contradict us when
we make assertions about the external world, unless
there is some logical inconsistency in our assertions.
The information we give to the computer will be taken
as fact, and if we specify the context it will be
recorded as fact in a particular context. The computer
only needs to be able to determine what conceptual structures
we had in mind when creating a communication, and
consult the specifications of its program to determine
what decisions need to be made based on those conceptual
structures. So, for this purpose, yes, there are physical
forms that are referenced by these conceptual structures, but
the non-robotic computer has no possible way of contradicting
the models and conceptual structures in the mind of the
communicators, so the philosophical question of the
relation of those structures to those physical entities they
represent does not arise in our communication programs.
As I mentioned, our goals at this point are very
practical. But the ability to encode and decode
conceptual structures in the form of a communication
will not be in any way contradictory to the goals
of a robotic/sensor system that may need to create
and verify internal models of the external world. The
two can interact when the time to do so is at hand.
For our present purposes, we trust that the person
encoding the model of the external world has done a
competent job, for whatever purposes the encoding
was done (which purpose will also be part of
the encoding). Whether the program works as intended
is the only way I know of to determine if the
encoding was done properly. (01)
Pat (02)
Mike Brenner wrote:
> I would like to become familiar with ontology tools
> capable of expressing more than simple "literal" meaning.
>
> I would like a "true meaning", by which I mean
> a clustering of information showing how the literal meaning
> maps to the multiple contexts. That mapping
> includes constraints, dependencies, and effects
> from partially defined and partially related chains of symbols.
>
> Thus, I don't see meaning as related to conceptual forms in
> the mind of the reader, but rather to physical forms which
> are the context in which the communication takes place.
>
> Mike Brenner
>
>
--
=============================================
Patrick Cassidy (03)
MICRA, Inc. || (908) 561-3416
735 Belvidere Ave. || (908) 668-5252 (if no answer above)
Plainfield, NJ 07062-2054 (04)
internet: cassidy@xxxxxxxxx
============================================= (05)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Unsubscribe/Config:
http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (06)
|