[Top] [All Lists]

Re: [ontolog-forum] Re: Semantics

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Adam Pease <adampease@xxxxxxxxxxxxx>
Date: Tue, 03 May 2005 07:55:06 -0700
Message-id: <>
Hi Duane,
   A good way to check this issue I think is to look at SUMO.  The only 
"primitives" that are undefined in the ontology are the logical operators, 
which are defined in the SUO-KIF language specification.  Every term builds 
up definitions from the logical operators, and then other terms.  I don't 
believe there are any circular definitions in the ontology, although I'd be 
glad to discuss any that appear so, since that would be a bug.
   Ron Schuldt brings up the issue of language, which is often a problem 
with informal ontologies that do not separate logical terms from linguistic 
tokens.  A formal ontology such as SUMO must distinguish language from 
logic to avoid the sort of confusion or imposition he describes.  SUMO does 
this by keeping WordNet synsets and SUMO terms related but separate.
   Lisa brings up an interesting issue about "semantic shift".  In a formal 
ontology, terms mean exactly what the axioms say they mean.  However, very 
few terms in the common sense world have necessary and sufficient 
definitions.  Each new axiom in which a term appears adds to its formal 
definition.  It is quite possible to use a term in a way that is consistent 
with its semantics so far, and then extends those semantics.  If there are 
a lot of terms in an ontology, the engineer may make an error in the use of 
the term.  Testing for logical contradictions or against a regression test 
suite can help.  Because common sense terms don't often have sufficient (in 
the mathematical sense) definitions they are unfinished products.  The 
intent of the engineer who created the term may be different from that of a 
later engineer who wants to add semantics to the term.  I don't see any 
general solution to this other than controlling the upper ontology to a 
degree by having a single architect, and well trained 
collaborators.  Fortunately, that's the case for SUMO.    (01)

Adam    (02)

At 07:25 AM 5/3/2005, Duane Nickull wrote:    (03)

>Chris Menzel wrote:
>>Well, if a definition of a concept *does* make use of concepts that are
>>axiomatized in terms of concept being defined, then it is just a bad
>This is what I wanted to explore.  Look at the english dictionary - it 
>uses all the words that are defined in the dictionary to define the words 
>in the dictionary. A great deal of care is taken to avoid direct inclusion 
>of terms in circular references however most of the words defined by the 
>dictionary are probably used in definitions of other words.
>Is that really an ontology?  Are there formulas that state the number of 
>levels a word must be reasonably not used in a set of definitions until it 
>is used again?
>1. A Company is a military unit, typically consisting of 100-200 soldiers
>2. A Battalion is an army unit usually consisting of a headquarters and 
>three or more companies
>3. A Division is an military unit large enough to sustain combat
>4. A Regiment is a military unit, larger than a company and smaller than a 
>In the definition of Regiment #4, we have used words to explain it that 
>were just defined themselves #1,3.  #2 is superfluous yet aids in 
>providing semantics to some degree (or does it?).
>Sorry to once again be the loose cannon ;-)
>Senior Standards Strategist - Adobe Systems, Inc. - http://www.adobe.com
>Vice Chair - UN/CEFACT Bureau Plenary - http://www.unece.org/cefact/
>Adobe Enterprise Developer Resources  - 
>Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>Shared Files: http://ontolog.cim3.net/file/
>Community Wiki: http://ontolog.cim3.net/wiki/ To Post: 
>mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (04)

Adam Pease
http://www.ontologyportal.org - Free ontologies and tools    (05)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (06)

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