[Top] [All Lists]

Re: [ontolog-forum] Ontologiesassocialmediators(was:Ontologydevelopment

To: edbark@xxxxxxxx, "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Adrian Walker <adriandwalker@xxxxxxxxx>
Date: Fri, 4 Dec 2009 13:35:00 -0500
Message-id: <1e89d6a40912041035p3aacf011o26044123fe2045b6@xxxxxxxxxxxxxx>
Hi Ed,

You wrote

 ...A concept is defined by reference to its properties or by reference to its
instances.  In many cases, the reference to instances is exemplary, and
intended to be inductive.  Precisely because these lists (properties and
instances) are open and grow with experience, the process is clumsy and
does not always produce "perfect" alignment of the concepts.

There's actually another way to define a concept, one that to some extent bridges the person-machine gap.  You can define a concept in terms of other concepts.  For example:
we need to reduce projected world yearly emissions of CO2 by some-number million metric tons per year by 2025
increasing some-activity by some-amount percent would reduce world CO2 emissions by 1 million Metric Tons/yr
multiplying that-number by that-amount and rounding to 1 place gives some-quantity
increasing only that-activity by that-quantity percent by 2025 would meet world 2005 CO2 emission requirements
In the example, the concept expressed by the last line is defined in terms of the concepts in the first three lines.

So, one might ask, is there infinite regress here?  Actually no, because one can ground a concept of this form by using it as a heading for a table of data [1,2].

                                        -- Adrian

[1]  www.reengineeringllc.com/demo_agents/CarbonDioxideEmissons1.agent

[2]  www.reengineeringllc.com/A_Wiki_for_Business_Rules_in_Open_Vocabulary_Executable_English.pdf


On Thu, Dec 3, 2009 at 3:07 PM, Ed Barkmeyer <edbark@xxxxxxxx> wrote:

> In order to be able to agree on any piece of knowledge we must establish the identity of the subject of discourse and align our own concepts against each other.

Yes.  But we cannot expect such alignment to be perfect in any sense.

> Such an alignment is not possible by using object, properties and relations and events as identifiers. Even if we agree that a concept is an objectand an object is a totality of propeerties and a list of real life items displaying/materialising those properties (usually also used as examples). These lists must be open as they grow with the advance of learning.

And yet, this is exactly how people have always succeeded in
communicating concepts, and assigning shared terminology.  A concept is
defined by reference to its properties or by reference to its
instances.  In many cases, the reference to instances is exemplary, and
intended to be inductive.  Precisely because these lists (properties and
instances) are open and grow with experience, the process is clumsy and
does not always produce "perfect" alignment of the concepts.

But like human communication, no claim is made that ontologies produce
perfect alignment of the intent, either.  What is critical is that it
produces practical alignment -- the term serves the purpose of the
interpersonal communication; the ontology is reliable with respect to
the properties cited.

> Because  the process of increasing our knowledge is not recorded, the path is important for the individual and the community.

This is true.  That is, the concept that any person has, or any
community shares, is formed by experience, and will reflect that
experience to some extent.  And the varying ability of humans to
abstract and analogize means that different individuals may acquire
different concepts from the same experience.

["Some people can learn from the mistakes of others; some can learn only
from their own mistakes; and some don't seem to learn from either."  --
Mark Twain]

> resolution rather than controversion is required.

That is simply not true.  Both are required.  We learn from comparison
and discrimination, but also from counterexamples and false analogies.
"All that glitters is not gold."

> But from the very beginning of schooling we pass on knowledge in school subject frames without harmonizing definitions or cross-referencing.

Yes, but no one's education stops there.  At some time, commonly
supposed to be around age 12, children become capable of critical
thinking.  And thereafter, in some curricula, we try to teach them the
skills of argument, comparision, unification, etc., in the hope that
this will lead them to harmonization and cross-reference.  But those are
skills and they depend in part on individual talent.  Just as no amount
of coaching can make an average striker "bend it like Beckham", no
amount of teaching can create an Einstein.

> Today you have concordance programs and it would be a simple job to find out any inconsistencies in usage, adpect or approach.

I hope no one believes this.  The equivalence of the concepts in the
heads of two authors is unknowable.  Some level of practical equivalence
may be gleaned from their writing.

This is also where the problem of the culture of speech communities
enters in.  A frequent phenomenon in government and large corporations
is management writing that is intended for upper management and is
confusing or contradictory for the people actually doing the work.  The
communities use the same catch phrases for the same subject matter, but
they are used at different levels of abstraction and with different
intent.  A concordance, however, would likely identify them as having
the same meaning, because the environmental terminology matches, by
(intentionally) having the same kind of differences in interpretation.
Think of "Service-oriented architecture".

> Therefore the representation of knowledge should not be static as in abstract networks, but dynamic, thus procedural, which means  that time and space must be included in an ontology together with verbs that represent realtions at a better detail (for the sake of identification) than the current relations you know well enough are insufficient.

I may be missing Ferenc's point here.

This may be an argument for what should be in a physical ontology, but,
to follow an earlier line, I cannot imagine how to apply spatial notions
to most concepts in linguistics or psychology or even the famous
oenology ontology.  And the application of temporal notions to those
fields reflects specific dynamics that are concepts in those fields,
such as "phonological/consonantal drift" and "aging".  Dynamic
properties are just properties.  Concepts are verb-like and noun-like
and modifier-like; some of the verb-like concepts are "dynamic" in
nature.  "procedural" is an even narrower concept.

The dynamic behavior of a body of knowledge is a different concern from
"dynamic concepts".  As Ferenc says, the sum of all human knowledge is
constantly growing.  The question is:  How does that relate to knowledge

The representation of knowledge in any captured form, including
ontologies, is by definition static.  It only changes by replacement.
Its longevity and value is dependent on the breadth of experience that
was represented in the encoded knowledge itself, and the relative rate
of evolution and revolution of all knowledge in that domain.  Isaac
Newton's physics was accepted for 225 years; Noam Chomsky's linguistics
was accepted for about 30.  And both are still respected, but they are
now understood as being valid only for a certain set of uses and in
certain views.  Nothing we do about knowledge engineering as we
understand it can change this.  When there is more knowledge in an area,
some elements of some commonly used texts and ontologies will be
obsolete and perhaps invalid, or like Newtonian physics, no longer valid
for all uses, but still valid for many of the same old kinds of uses,
including most new instances of them.


Edward J. Barkmeyer                        Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263                FAX: +1 301-975-4694

"The opinions expressed above do not reflect consensus of NIST,
 and have not been reviewed by any Government authority."

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
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (01)

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