I generally agree with John
Bateman's contextual discussion on goal and functions affecting what
boundary means (“ontological characterizations.” As John described it). The experimental work on robot-human
interactions in spatial settings provides real data as I suggested in my earlier
message drawing on epigenetic robotics.
I think that his
statement
>"It took us three hours to cross the border."…..Sounds pretty
thick to me.
suggests how linguistic data gives
very useful indications. It’s thru
our own mental models which we use in turn for
> modelling decisions
relevant for building
>ontologies, but only when listened
to.
There is a degree of commonality to
our own experience based mental models as designers of ontological models from
insights offered by linguistic data we have to be aware that our lifelong
developed mental models will differ.
That’s one reason that letting robots develop their own models rather
than building them in offers some check on unconscious bias. Of course we have realize that our
general design principles for developing cognition may also be flawed and have
to be reformulated.
Gary
Berg-Cross
-----Original Message-----
From:
owner-standard-upper-ontology@ieee.org on behalf of John A. Bateman
Sent: Thu 7/14/2005 7:57 AM
To: Murray Altheim
Cc: sowa@BESTWEB.NET; standard-upper-ontology@IEEE.ORG
Subject: Re: CG: Re: Re: Whole and Parts (and
boundaries)
> I have no real data about people talking about
boundaries,
> but I have some info about people talking about walls
--
> i.e., whether a wall is part of a room or a boundary
>
between rooms or whether part of the wall is part of the
> room or part
of the wall is part of the boundary.
>
> If you ask them, people
have an incredible number of
> different thoughts ranging from
confusion, to irritation,
> to lengthy and irrelevant discourses on the
topic -- and
> essentially all of those responses are useless for
any
> serious kind of application.
These kinds of issues are
ABSOLUTELY ESSENTIAL for
any one to make any statements about 'how
people
do things'. As John S. says, if you haven't collected
(or have
access to) the data where people are actually
going about tasks trying to
solve problems involving
the concepts, then it is better to say
nothing.
In my own group, putting people in situations where
they
have to talk about and solve spatial problems is
the fundamental
methodology without which we say
nothing about 'how people' do stuff
(http://www.sfbtr8.uni-bremen.de/i1).
On the basis
of these experiments (usually robot-human
interactions in
spatial settings),
we propose ontological characterisations.
Absolutely
obvious, is that characterisations depend
on goal and functions. So even
the word 'boundary'
is not yet sufficiently defined to say much that
is
sensible, but as soon as it is used in context,
that complexity usually
vanishes.
Example:
> For example, when lay people (as opposed to
geographers
>>or politicians) talk about the boundary between
two
>>countries, they generally don't think about the
thickness
>>of the boundary.
Absolute nonsense.
Data:
"It took us three hours to cross the border."
Sounds
pretty thick to me.
As soon as one then attempts to 'explain away'
the
usage (e.g., "well, they were not strictly talking about
the border
but about the road up to the crossing point,
which is actually a line of no
thickness"), I think
the point has been missed. Either you use
the
linguistic data or not. If not, fine, then we
can go with boundaries
defined mereotopologically,
fiat or otherwise). If linguistic data
('how
people think about...'),
then it should be done properly. That's my
only
point here I think. I believe the
linguistic data do give very useful
indications
of modelling decisions relevant for building
ontologies, but
only when listened to.
John B.