[om-list] Facet based classification

[Thomas L. Packer at home]

I like what Mark has said.

tomp

~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Víðar sum quem nihil obstat.
www.Ontolog.Com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~


----- Original Message -----
From: "Mark Butler" <butlerm at middle.net>
To: <om-list at onemodel.org>
Sent: Monday, 02 September, 2002 10:04
Subject: [om-list] Facet based classification


> Hello everybody,
>
>   I fundamentally agree with the idea behind "facet based" classification.
A strict hierarchical classification system (i.e. forced single inheritance)
is hopelessly inflexible for any comphrensive model of knowledge.  Any real
world object may be classified by an arbitrary number of dimensions, and
any normative hierarchical classification system applies a fixed number of
them in a fixed order to any given object.
>
> Tom has developed a model that is very useful.  Any given object or
abstraction may be thought to occupy a point, or more generally a region, in
a multi-dimensional space where each attribute of an object gives it a
position or extent in one or more corresponding dimensions.
>
> The most obvious dimensions are space and time.  Position and extent in
space and time are obvious attributes of every real world object.  Most
classifications are based on much more subtle dimensions, however.  For
example, the plant-ness or animal-ness of a living organism.  While most
existing examples fall clearly on one side or the other, I dare say there is
no definition of a plant that has no ambiguity with the corresponding
definition of an animal, and vice versa.
>
> Now of course, as a matter of philosophical realism, neither plants nor
animals care one way or the other. Plant-ness or animal-ness is a mental
distinction we have adopted that can't be said to be *real*, i.e. to have an
independent existence outside of our own minds, even though it clues us in
to properties that are real, such as mobility and intelligence.
>
> There lies the death of normative my-way-or-the-highway hierarchical
classification systems - basically they are mental conveniences that often
tell us more about one way of thinking about reality than about reality
itself.  Now while modeling beliefs about reality is unavoidable, great care
should be taken not to "hard-code" one belief system or world view into a
comprehensive model of knowledge, largely because all mortal world views, or
schemas, are strictly works in progress, even if most aspects are known to a
high degree of certainty.
>
> Tom and I would like to see a system with analytical capabilities as well
as representational capabilities.  A very simple example would be to
determine given an organism with various properties, how plant-like or
animal-like it is.  That requires a logical model of the organism in
question and a way to relate it to logical models of what it means to be a
plant or an animal.  Dimensional analysis seems to be a good general purpose
way to get started.
>
> A classic complex example is determining to what degree one event was
*caused* by another event. I.e. did the assassination of the Archduke of
Austria cause World War I, or was it largely incidental to a dispute ready
to boil over at any decent provocation?  Of course, that is an exceedingly
complex question that requires an enormous amount of evidence to come to a
reasonable conclusion about, but variations on the same theme are a topic of
major academic and practical interest.
>
> Fault detection and classification in everything from car engines to a
complex industrial processes is a field of enormous practical importance
where the necessary information for rational conclusions is much more
readily available.
>
> On the more theoretical side, in rigorous disciplines like mathematics, it
is reasonably straightforward to construct simple theorem provers as well as
proof checkers.  Most logical arguments are considerably more fuzzy than
mathematical proofs, but I believe that there is a reasonable basis for
analyzing well ordered bodies of knowledge according to time honored
principles of logic, probability, and statistics, to the degree that such a
system can help identify and demonstrate major logical flaws in (or at least
the unmentioned premises behind) much of what passes for knowledge these
days.  Dimensional analysis functions as an effective way to transform
normal natural language definitions and statements into a model suitable for
such calculations.
>
>  - Mark
>
> _______________________________________________
> om-list mailing list
> om-list at onemodel.org
> http://www.pairlist.net/mailman/listinfo/om-list