[Top] [All Lists]

Re: [ontolog-forum] Levels

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Wed, 14 Feb 2007 21:05:43 -0500
Message-id: <45D3BFF7.6040701@xxxxxxxxxxx>
Kathy and Bill,    (01)

I strongly sympathize with the requirement for probabilities,
and if we provide a format that supports entrenchment, I
believe that probabilistic level(s) should also be included.    (02)

KBL> Probabilistic information ought in some cases to be extremely
 > high up in a "levels of entrenchment" ordering.    (03)

The level of "entrenchment" has nothing to do with level of
importance.  I would agree that a probability statement may be
very important.  But an observation that "p is true" would
overrule a statement that p has a 95% probability of being false.    (04)

KBL> In many cases, there just plain isn't any such thing as necessary
 > and sufficient conditions.  Ask any expert in machine learning or
 > classification.    (05)

I agree.    (06)

KBL> But in natural systems, necessary and sufficient conditions are
 > more the exception than the rule.    (07)

I definitely agree.  But there may still be a few necessary statements,
such as "every dog is an animal", even though there may be no clear
set of sufficient conditions for distinguishing a dog from a wolf or
many other kinds of animals.    (08)

KBL> Imports are fundamentally different from cut-and-paste.  That's
 > because in an open world environment, things change.  The specs
 > tend to ignore this issue, but it's very real.    (09)

I agree.  In previous discussions, I have advocated tools such as
a metadata registry for registering ontologies and recording the
history of who, what, when, where, why, and how they were written,
modified, updated, etc.  Many people have thought long and hard
about those issues, and we should take advantage of their work.    (010)

BA> If "ontologies" are supposed to include sophisticated mechanisms
 > for handling probability distributions and are supposed to support
 > classification, then what would be the role of epistemology and
 > phenomenology in such ontologies?    (011)

The mechanisms for handling probability would not be stored in the
metadata registry.  Nor would we put a theorem prover there.  What
would be included in the text of an ontology is a set of statements,
plus whatever appropriate metadata we agree should be there.    (012)

The question of who made a particular statement and its justification
could be stored in a metadata registry as background information.
But it would not be in the text of the ontology itself.  The source
and method of discovery would be outside a statement of probability
just as it would for a statement in first-order logic.    (013)

BA>  Is it the case that you reject the traditional divides between
 > ontology, epistemology, phenomenology, and semantics in natural
 > languages?    (014)

I personally believe that the subdivisions of academic departments
are often irrelevant to some kinds of practical work.  But that is
independent of whether we use FOL or probability.    (015)

John    (016)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (017)

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