(398) previous ~ index ~ next
To: TDT List <tdt-distrib@unagi.cis.upenn.edu>
From: Jonathan Fiscus <jonathan.fiscus@nist.gov>
Subject: [Fwd: [Fwd: TDT4 Req'd conditions]]
Date: Fri, 20 Sep 2002 15:15:09 -0400
Folks,
I incorrectly had automatic story boundaries for the NED task... It's
fixed below.
Jon
Story segmentation
------------------
Source language English, Mandarin and Arabic
Source data type ASR transcription
Maximum decision deferral period
10,000 words for English and Arabic
15,000 characters for Mandarin
All of these index files: seg_SR=bnasr_TE=arb,nat.ndx
seg_SR=bnasr_TE=man,nat.ndx
seg_SR=bnasr_TE=eng,nat.ndx
Topic Tracking
--------------
Basic Required Conditions
Topic training language English
Source language All
On-topic training stories 1
Off-topic training stories 0
Source data type text sources and manual transcription
of audio sources
Story boundaries Reference boundaries
One of these index files: trk_SR=nwt+bnman_TR=eng_TE=mul,eng_Nt=1.ctl
trk_SR=nwt+bnman_TR=eng_TE=mul,nat_Nt=1.ctl
Alternate ("Challenge") Conditions
Topic training language English
Source language All
On-topic training stories 4
Off-topic training stories 2 and 0
Source data type text sources and ASR transcription
of audio sources
Story boundaries Reference boundaries
One of these index files: trk_SR=nwt+bnasr_TR=eng_TE=mul,eng_Nt=4.ctl
trk_SR=nwt+bnasr_TR=eng_TE=mul,nat_Nt=4.ctl
New Event Detection
-------------------
Source Language English only
Source data type Text sources and the TDT4
automatic transcription of audio sources
Maximum decision deferral period
10 source files
Story boundaries Refernce boundaries
Index File: fsd_SR=nwt+bnasr_TE=eng,nat.ndx
Topic Detection
---------------
Source Language English, Mandarin and Arabic
Source data type Text sources and the TDT4
automatic transcription of audio sources
Maximum decision deferral period
10 source files
Story boundaries Reference boundaries
One of these index files: det_SR=nwt+bnasr_TE=mul,eng.ndx
det_SR=nwt+bnasr_TE=mul,nat.ndx
Link Detection
--------------
Source Language English, Mandarin and Arabic
Source data type Text sources and the ASR transcription
of audio sources
Maximum decision deferral period
10 source files
Story boundaries Reference boundaries
One of these index files: lnk_SR=nwt+bnasr_TE=mul,eng.ndx
lnk_SR=nwt+bnasr_TE=mul,nat.ndx
-------------------------------------------------------------
To unsubscribe from tdt-distrib, email majordomo@ldc.upenn.edu
with "unsubscribe tdt-distrib" in the body of the message.
(398) previous ~ index ~ next
Last updated Mon Nov 11 14:16:27 2002