Tom Gruber <gruber@eit.com>
Message-id: <ab15467f090210043f98@[]>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 14 Dec 1994 17:21:02 -0800
To: clucas@nas.nasa.gov (Jean E. Clucas)
From: Tom Gruber <gruber@eit.com>
Subject: what-is-an-ontology
Cc: srkb@cs.umbc.edu
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
>I'm curious. Did you coin this use of the word ontology, and if so,
>why? If not, do you know who did, or what that person's
>reasoning was, in choosing to give the word ontology a new meaning?
>Sorry if this sounds like a flame, I am really just curious.

No I didn't coin it.  The word has been used by AI people for as long as I
can remember to refer to a class hierarchy (and related things like data
dictionaries).  It isn't *that* far out of line if you consider that we are
talking about the categories of existence for an artificial agent, a
perspective the classical philosophy on ontology doesn't usually consider
(and the modern philosophy on "Formal Ontology" looks an awful lot like
what you see in the commonsense sections of KR conferences).

My definition merely reflects the progression of the AI use of the word to
account for a generalization of the specification language (class hierarcy
becomes axiomatization of vocabulary or any other knowledge-level
specification), an admittedly AI-ish view on knowledge (knowledge as
attributed -- Newell's knowledge level paper), and the pragmatics of
ontologies in knowledge sharing among programs (ontology as specification
and as agent communication language).

To hunt down an early culprit, look in the old papers about terminological
reasoning or maybe knowledge representation for NLP work.  I've CC'd the
relevant discussion list; perhaps someone can put the finger on who first
coopted the term "ontology" for AI.