(024) previous ~ index ~ next
To: Hubert Jin <hjin@bbn.com>
From: "George R. Doddington" <doddington@sri.com>
Subject: Re: DET curve of TDT1
Date: Mon, 13 Apr 1998 08:24:19 -0400
> We are testing our system on TDT1 (tracking), and would like to know
> how to score the events that do not have enough stories in training.
>
> For example, event #7 only has two stories with 'yes' labeled. Do people
> include the result (in case of N = 16) for that event in the DET curve?
> (Or just ingore the events that do not have the minimum stories for
> training since it implies all the testing stories are 'no' by default?)
Official TDT1 scoring for tracking included only those events for which
there were at least 16 stories.
--
George R. Doddington [at home]. Reply to doddington@sri.com,
or call 703/556-3434 [at home], or 301/688-0976 [at NSA]
(024) previous ~ index ~ next
Last updated Wed Sep 9 09:40:47 1998