(150) previous ~ index ~ next
To: Rich Schwartz <schwartz@bbn.com>
From: George Doddington <doddington@nist.gov>
Subject: Re: Comments on the TDT3 performance metric
Date: Fri, 23 Jul 1999 15:51:09 -0400
I have the pleasant feeling that the dust is beginning to settle on
this issue. So I really shouldn't respond to your note..... But,
I feel compelled to comment on the first remark that you made:
> 1. The cost model tries to model an imagined application rather than
> a technology. This is somewhat in contradiction to your general
> stated goal of trying to measure performance a little bit more
> generally oriented to the technology.
Well, yes and no. Yes, the a priori target probability factors into
the cost model. I.e., the richness of the source has a direct effect
on performance. However, this is an application parameter and is not
a (random) function of test corpus statistics. So, for example, the
size of a topic in the test corpus will have no influence, per se,
on the performance measure.
--
George Doddington at NIST: doddington@nist.gov or 301/975-3261
(150) previous ~ index ~ next
Last updated Thu Aug 19 16:14:47 1999