W3C home > Mailing lists > Public > www-webont-wg@w3.org > July 2002

Chris' issue: properties and relations

From: Christopher Welty <welty@us.ibm.com>
Date: Fri, 12 Jul 2002 09:58:55 -0400
To: www-webont-wg@w3.org
Message-ID: <OFAF017E8E.C12E7064-ON85256BF4.004A673D@pok.ibm.com>

This is bothering me so much I decided to try and write up a proposal for 
a new issue.  I don't know the proper process for that, but here's a 
description of the issue anyway:

ISSUE: properties in OWL are relations so lets call them that

In logic and mathematics, a "property" is defined to be a unary predicate. 
 A "relation" is defined to be an n-ary predicate where n>1.  A "binary 
relation" is defined to be a binary predicate.  I will limit myself to 
using the term "relation" and "binary relation" synonymously here, since 
OWL only deals with binary relations.

The notion of a daml:property corresponds to a logical relation.  The 
notion of a daml:class corresponds to a logical property.  Why is there 
this confusion???

The confusion stems from the "frame-based" treatment of complex objects as 
sets of object-attribute-value triples with the object fixed.  I.e. we 
like to talk about an object Chris as being:

Chris::
  has-father: Bill

Here, "has-father" is a relation s.t. the pair <Chris,Bill> is in the 
extension.

However, when we talk about Chris' father, mathematically and logically 
THIS IS A PROPERTY.  It is the property of "being Chris' father": 
father-of-chris(Bill).  For any relation, if you fix the domain to a 
single element, you have a property.  But note that it is not a property 
of Chris, it is a property of Chris' father.  So when I examine a "frame" 
for its "slots", I am not looking at the "properties of that frame".

Another subtle confusion here is that "having a father" is also a 
property.  I could say has-father(Chris) and intend it to mean that "it is 
true that Chris has a father". 

I honestly believe that using "property" to mean "relation" is the worst 
possible choice.  I always complained about using "role" in description 
logics, but "property" is one of the few things I can think of that is 
worse.




Dr. Christopher A. Welty, Knowledge Structures Group
IBM T.J. Watson Research Center
PO Box 704, Yorktown Heights, NY 10598, USA
+1-914-784-7055 Fax: +1-914-784-6078
Received on Friday, 12 July 2002 09:59:26 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:51 GMT