Re: Types, sets, and relations

Message-id: <>
Date: Tue, 27 Apr 1993 14:44:29 -0800
To: sowa <>,,
From: macgregor@ISI.EDU
Subject: Re: Types, sets, and relations
I am complete agreement with the basis for John's desire to distinguish
types, sets, and properties:

>Bottom line:  KIF has a purely extensional semantics that eliminates
>any distinction between types, sets, and monadic relations. ...
>However, there are other systems that do make such distinctions on
>various grounds -- formal, pragmatic, esthetic, or traditional.
>Some of those distinctions turn out to be very important when dealing
>with modal and intensional issues.

What I disagree with is John's proposed solution, which is to encode
the distinction within syntactic forms which would be officially
(at least for the near future) to be semantically equivalent.

At present, KIF omits some things, such as types and roles, that some
of us regard as fundamental to a KR system.  Calling types an unresolved
research issue is, I believe, significantly overstating the problem.
"sorts" have long been distinguished from monadic predicates in many
mathematically-oriented logics.  My impression is that the principal
controversy surrounding the issue of "Type" as a KIF entity is that
Mike Genesereth doesn't like it.  At least, I have yet to hear anyone
else speak out against inclusion of types in KIF.  Thus, its possible
that if the Interlingua committee met to vote on the issue (say at
the upcoming AAAI conference) we might discover that there was a large
amount of agreement in favor of distinguishing types from properties.
Such a result, if it occurred, would be preferable to waiting an 
indeterminate number of years using syntactic conventions to disguise
an epistemological hole in KIF.

- Bob

Robert M. MacGregor                           
USC/ISI, 4676 Admiralty Way, Marina del Rey, CA 90292      (310) 822-1511