Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: Call for vote on SUO Scope and Purpose



At 07:01 PM 7/27/2000 -0700, WBurkett@pdit.com wrote:
Adam:
 
>  I'm puzzled by some of your comments.  Comments below:
You don't seem puzzled!  You just seem to hold a different set of opinions. :-)
Bill,
  No, I was genuinely puzzled about why you were apparently arguing for formal definitions in logic as the basis for the SUO when that's just what is intended.  We're in agreement on that point.  I guess the statement of purpose just wasn't clear in that regard.
My comment to your below:
 
-----Original Message-----
From: apease [mailto:apease@teknowledge.com]
Sent: Wednesday, July 26, 2000 5:49 PM
To: WBurkett@pdit.com; 'Schoening, James R CECOM DCSC4I'
Cc: SUO (E-mail)
Subject: RE: Call for vote on SUO Scope and Purpose

Bill,
  I'm puzzled by some of your comments.  Comments below:

At 05:34 PM 7/26/2000 -0700, WBurkett@pdit.com wrote:

<snip>

(1) I believe that the objective of 1000-2500 terms is far too large to be practical.  In my experience, I've found that the practical upper limit on the number of independent object/entity types in a schema is ~300 types.  Models larger than this simply cannot be comprehensively understood by a single person.  (One may argue that it's possible to understand larger models with the aid of a meta-level organizing structure - and I agree - but then *this* structure becomes the upper level ontology.) (And there is the further question of criteria for selecting and discriminating between ontological elements.)

Folks in the HPKB project including Cycorp, Teknowledge and Stanford used the Cyc upper model of 3000 terms successfully in several tests.  I take this as an existence proof that refutes your assertion, or more gently, maybe it just refutes the need for an individual to comprehend the entirety of a model for that model to be useful.

[William Burkett] But aren't these people professionals in this domain that have a great deal of experience and education in the craft of developing (for want of a better term) knowledge models?  And tons of experience with that one particular model?  How is Joe Schlub Java Programmer have any hope of understanding the SUO with 3000 items in it?  I still maintain that Mr. Everyday Schema developer will never develop or use a model this large (and probably would never want to.)
 
So, if we're serious about application developers using this SUO to support application interoperability, it had better be simple and comprehendable.
I believe that the SUO will in fact just be a start as to what is needed for having shared conceptualizations of terms.  If we have a smaller model with say only 300 terms in it there won't be enough shared semantics for two systems to be able to communicate.  For example, take the Cyc public upper ontology of #$temporallIntersects.  The semantics are very limited.  Now travel down the hierarchy of predicates in turn to #$actors, #$doneBy and #$performedBy.  If we needed to talk about a manager giving direction to an employee or a customer giving an order to a clerk, just saying that they intersect temporally is not enough.  Just having the #$performedBy relation without the notion that they intersect temporally could lead to some erroneous conclusions.
(2) I think the SUO should be specified in a formal language to facilitate data processing.  I don't think terms and definitions are enough for practical applications.

Specification in a formal language is precisely what is intended "..An ontology is a set of terms and formal definitions..."  I've proposed KIF or some derivative thereof which is a formal language.

[William Burkett] Doing it in KIF would be fine by me, but my reading of the Scope and Purpose made it sound like the SUO was a dictionary.  There was no mention of formal languages for declaring the ontology.  ("formal definition" <> "formal language" - it could just be legalized English.)

Ok, maybe the new version of the Scope and Purpose that incorporates everyone's comments from the vote can clear this up.

Adam

Bill
 

-----------------
Adam Pease
Teknowledge
(650) 424-0500 x571