[Top] [All Lists]

Re: [ontolog-forum] Universal Basic Semantic Structures

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Wed, 26 Sep 2012 16:00:36 -0400
Message-id: <50635EE4.4000903@xxxxxxxxxxx>
Andries, Doug, William, and Rich,    (01)

Any general purpose ontology has to be useful for the full range
of applications for computer systems.  That means it must be span
the range of semantics from informal language to every branch of
science, technology, business, law, medicine, and the arts.    (02)

The most successful approaches span that range by using an
underspecified upper level ontology with very few axioms
and with specialized microtheories for various kinds of
applications.    (03)

> Design of buildings usually starts with the design of spaces
> (rooms etc.), then the walls are added. I consider the spaces
> to be components of the building (not as an assembly, but
> as a component).    (04)

That's a useful view by architects, and it should be supported.
But geometry is the general theory of space, independent of any
content or physical boundaries.  The architects who design
buildings and the contractors who build them use geometry that
is consistent with the mathematicians and physicists.    (05)

> I question whether a physical area is by definition two dimensional.
> Mathematical area's are two dimensional. But two dimensional area's
> in physical reality seem to be abstractions.    (06)

Of course it's an abstraction.  Physicists, engineers, and building
contractors are well aware of the approximations needed in their work.
But they use classical Euclidean geometry to measure anything on planet
earth that's not moving close to the speed of light.  And they have
established "best practices" for dealing with errors in measurement.    (07)

> This is related to the concept of 'surface'. A surface can have a roughness,
> a color, a hardness, a temperature, a strength, etc. I think that it can't
> have such properties when it would be only two dimensional.    (08)

Yes.  But countries and school districts specify their boundaries by
using classical Euclidean geometry in polar coordinates (latitude and
longitude).  They don't specify roughness or depth until a specific
application needs to address those issues.  Then they switch to an
appropriate microtheory that deals with them.    (09)

> In some cases (e.g. mines, reservoirs) the subdivision of the earth
> requires an explicit third dimension.    (010)

Yes.  But the specification of the region begins with the latitude
and longitude.  The depth depends on the application.  Engineers
that specialize in mining or reservoirs will use very different
microtheories for dealing with the depth.    (011)

> And some, such as lakes and mountains, don't need a role.    (012)

Nature is independent of the way we think or talk about it.  But
people choose different ways of classifying natural phenomena for
different purposes.    (013)

The reasons for a classification must be recognized by an ontology.
And experts in different fields have different ways of classifying
the same features.  Consider a park ranger, a mountain climber,
a lumberjack, a mining engineer, or a real-estate developer.
They use very different microtheories.    (014)

> I find 314 specializations of #$touches...    (015)

Those and other examples from OpenCyc are important.    (016)

> Cyc made the distinction of treating geopolitical entities from
> geographical regions in 1998.  One could use a context in which
> the geopolitical entity was an agent, or one in which it was
> also a piece of land...
> I urge people who want to develop ontologies of common things to
> first check what OpenCyc has.  It can save a lot of reinventing of
> complex wheels.  Cyc is certainly far from perfect and has huge
> holes, but with hundreds of staff years being put into a publicly
> available ontology, many parts of which are very carefully considered,
> it can be used to speed up such ontology development projects.    (017)

I agree.  Nothing is perfect, but the Cyc developers invested
a thousand person-years of work by 2010.  Anybody who develops
a general purpose ontology should study the OpenCyc ontology
instead of re-inventing the wheel.    (018)

> Try to imagine a thing you could not classify at all. I just can't.    (019)

That's true.  Every act of perception classifies sensory input
in some way, and imagination builds on the memories of earlier
perceptions.    (020)

> no two people use exactly the same ontology, which is one of those
> things that make interpersonal communications so very faulty.    (021)

True.  But people are much more flexible than computers.  The major
challenge is to design more flexible computer systems.  We're not
going to achieve that if we force them to use a rigid, monolithic
classification that cannot grow and adapt.    (022)

John    (023)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (024)

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