Re: Connectedness & Togetherness

"Nicola Guarino" <>
Message-id: <>
Date: Mon, 23 Jan 95 11:01:55 +0000
From: "Nicola Guarino" <>
Subject: Re:  Connectedness & Togetherness
X-Mailer: VersaTerm Link v1.1
Precedence: bulk
Folks, I am forwarding this mail to SRKB since - after my previous posting
regarding the shift from the "ontology" list to skrb, I think there could be
interested people there. Still puzzled about the right place for these
discussions. Even more puzzled from John's points on contexts, especially points
1 and 5. 

-- Nicola

----- Forwarded message follows -----

Date: Thu, 19 Jan 1995 14:49:05 +0500
From: (John F. Sowa)
Subject: Re:  Connectedness & Togetherness
Message-id: <9501191949.AA06626@west>
Content-transfer-encoding: 7BIT
Content-length: 2766

Re contexts:  The notion of context, as I use it, is not a complex
notion at all.  Please do not confuse my technical term "context"
with any other usage that any one else has been using.  Let me say
a few words about what I mean by the notion of "context":
1. A context is any entity of any type whatever whose description
   has one or more "included" entities.  By "inclusion", I mean
  I mean a very general notion, whose specializations are
   "part-of", "contained-in", "attribute-of", "property-of", etc.
   This notion of "inclusion" is not a complex notion, but rather
   the intersection of all those other notions, e.g. "part-of", etc.
   I identify this "inclusion" relation with the basic primitive
   of mereology.  It corresponds to what many people call either
   "part-of" or "moment-of" -- see for example, Husserl's
   Logical Investigations and the kinds of things that people
   like Barry Smith, Peter Simons, etc., discuss about mereology.
   If the context happens to be a set, "inclusion" corresponds
   to subset-of.
2. My primitive notion of context has NONE of the various complex
   notions built into it, such as scoping, indexicals, abbreviated
   references, encapsulated objects & behaviors, import/export rules,
3. However, I very much sympathize with people like John McCarthy,
   who wants to use a single notion of context to solve many different
   problems of knowledge representation.  I claim that all of McCarthy's
   wishes can be satisfied by my very primitive notion of context
   together with the ability to state axioms about what kinds of
   operations are permissible on contexts of various types.
   None of those axioms are built into my notion of context.  If you
   have a context of undefined type T, no operations of any kind are
   AUTOMATICALLY defined for that context; but no operations of any
   kind are AUTOMATICALLY prohibited, if you wish to define them.
4. If you wish to define a theory that axiomatizes any kind of
   operations, properties, or whatever for a particular type of context,
   you may do so.  However, you must take full responsibility for any
   errors, contradictions, etc., caused by your axioms.  And no one
   else will ever be affected by your errors, unless they choose to
   adopt your axioms and definitions.
5. I claim that this notion of context is much more primitive than
   a definition that introduces "properties".  I choose to make
   properties objects (dependent objects in Husserl's terms).
   A property such as "color red" or a monadic predicate is-red can
   both be defined in terms of a context for an object that includes
   a dependent entity "red".
John John Sowa
(sorry about the extra "john", but I can't edit this file)

----- End of Forwarded message -----