Re: Evaluation [was Re: kif ] Robert Neches <neches@ISI.EDU>
To: "Matthew L. Ginsberg" <firstname.lastname@example.org>
Cc: interlingua@ISI.EDU, neches@ISI.EDU, email@example.com
Subject: Re: Evaluation [was Re: kif ]
In-reply-to: Your message of Fri, 17 Dec 1993 09:37:00 -0800.
Date: Fri, 17 Dec 93 12:11:46 PST
From: Robert Neches <neches@ISI.EDU>
> As a criteria for evaluating/revising KIF, I would like
> to see some focus on practical experience using it
> for translation purposes.
> You mean, I assume, that you would like to see this sort of
> information *IN ADDITION TO* conventional academic review of the work.
I see translation experiments as a stricter criterion, but actually
I'm raising a deeper issue:
For both academic review and empirical testing,
let's first clarify what questions we're asking.
Interlingua is intended to facilitate knowledge interchange, with
translation being one of the key means by which it could contribute.
Given that it's intended purpose is to facilitate knowledge
interchange, it seems worthwhile to evaluate/refine it by asking,
"how well does it facilitate knowledge interchange when we try it?"
I'm all in favor of any source of input that helps. Most of the
e-mail I've seen lately I interpret as debating whether KIF is
correctly principled. I just want to observe that this is not
identical to asking whether it's useful.
Experience has shown us that artifacts can be principled without
being useful and useful without being principled. (The ideal is
that it's both, of course.)
For the record, this is not taking a position on the quality of
the current version of KIF. Nor is it taking a position on
the relative importance of academic review vs. empirical testing.
This note represents a position on one issue only: what criteria
are used to evaluate an artifact like KIF.
I raise translation and empirical evaluation only because they
bear on one of my key criteria: utility.
USC/ISI, 4676 Admiralty Way, Marina del Rey, CA 90292 USA
Phone: 310/822-1511, Fax: 310/823-6714