Topic Detection and Tracking(TDT) 2 English Careful Transcription Text Corpus


Introduction

This file contains documentation on the Topic Detection and Tracking (TDT) 2 Careful Transcription Text Corpus, Linguistic Data Consortium (LDC) catalog number LDC2000T44 and isbn 1-58563-166-3. This corpus contains the transcription of the broadcast news audio file in the Topic Detection and Tracking (TDT) 2 Careful Transcription Audio Corpus, Linguistic Data Consortium (LDC) catalog number LDC2000S92 and isbn 1-58563-167-1.

Topic Detection and Tracking (TDT) refers to automatic techniques for finding topically related material in streams of data such as newswire and broadcast news. The TDT2 corpus was created to support three TDT2 tasks: find topically homogeneous sections (segmentation), detect the occurrence of new events (detection), and track the reoccurrence of old or new events (tracking). For further information on TDT2 please visit our TDT2 Information Pages.

Please see file.tbl for the directory structure of this publication, as well as a complete list of files.
The data files are located in the transcripts directory.

Data

The data files are transcripts of broadcasts of the copyrighted sources below from January through June 1998 and of Voice of America (VOA) news broadcasts from March through June 1998. These selected audio transcripts have had two or more passes since the release of a superset of this data in the TDT2 English Text, Version 2 to ensure that these transcripts have a higher level of accuracy.

Also, speaker identification has been added to the annotation with this release. Where the speakers could be identified they are listed by first name underscore middle name underscore last name (eg., Peter_Jennings and Mike_Von_Fremd). Where the speaker could not be identified, they were listed as source underscore 'anon' underscore sequence number (eg., ABC_anon_03 and PRI_anon_03). While efforts were made to ensure that no two speakers were tagged with the same speaker id there exists the possibility that a single speaker was tagged with two or more speaker ids (eg., ABC_anon_0x could be the same as ABC_anon_0y).

The SGML file tagging was verified via nsgmls on the utf.dtd.

LDC2000T44 may contain information on updates or other addenda.

Content Copyright

Portions © 1998 ABC
Portions © 1998 CNN
Portions © 1998 PRI


Contact: ldc@ldc.upenn.edu
© 1996-2000 Linguistic Data Consortium, University of Pennsylvania. All Rights Reserved.