2003 New Event Detection Timeline.Today is
13 May 2008 .
TDT group meetings are Mondays from 10-11 or from 10:30-11:30. If there is a talk scheduled at 11am, the meetings will start at 10.
2003 Summer Meetings:
TDT group meetings are Mondays from 2-3. Room CS303.
--
JamesAllan - 20 Mar 2003
TDT Corpus
- TDT-1 Corpus
- used for 1997 pilot-study
- 2 sources from Reuters and CNN
- About 16,000 news
- Covers July 1, 1994 to June 30, 1995
- 25 topics annotated by researchers
- TDT-2 Corpus
- used for 1998 test
- 6 English sources: New York Times, Associated Press Worldstream, CNN, ABC, PRI and VOA
- Covers from January 4 to June 30, 1998
- About 6,000 news
- 100 topics annotated by researchers
- TDT-3 Corpus
- used for 1999, 2000 and 2001 tests
- 8 English sources(New York Times, Associated Press Worldstream, CNN, ABC, PRI, VOA English, NBC and MSNBC) and 3 Mandarin sources (Xinhua News, Zaobao and VOA Mandarin)
- Covers from October 1998 to December 1998
- about 9130 news
- 60 topics annotated by researchers
- TDT-4 Corpus
- used for 2002 test
- 8 English sources, 7 Mandarin sources and 5 Arabic sources
- Covers from October 1 2001 to January 31 December 2002
- 90375 news
- 60 topics annotated by researchers
Annotation Guides:
Includes
rules of interpretation.
When looking for rules of interpretation, take a look at the
mapping between the old rules and the new rules.
Links to topic descriptions:
--
BinLiu - 25 Mar 2003
--
AlvaroBolivar - 30 Jun 2003
TASKS
to top