Reuters Transcribed Subset

Donated on 3/7/2008

This dataset is created by reading out 200 files from the 10 largest Reuters classes and using an Automatic Speech Recognition system to create corresponding transcriptions.

Dataset Characteristics

Text

Subject Area

Business

Associated Tasks

Classification

Feature Type

-

# Instances

200

# Features

-

Dataset Information

Additional Information

Data Characteristics: -------------------- This data was created by selecting 20 files each from the 10 largest classes in the Reuters-21578 collection (http://www.daviddlewis.com/resources/testcollections/reuters21578/). The files were read out by 3 Indian speakers and an Automatic Speech Recognition (ASR) system was used to generate the transcripts. More about the ASR system can be found in [1]. Such a dataset will be really helpful to study the effect of speech recognition noise on text mining algorithms. The first work which refered to this dataset was on noisy text classification[2]. Data Format: ---------- There are 10 directories labeled by the topic name. Each contains 20 files of transcriptions. References: ---------- [1] L. R. Bahl, S. Balakrishnan-Aiyer, J. Bellegarda, M. Franz, P. Gopalakrishnan, D. Nahamoo, M. Novak, M. Padmanabhan, M. Picheny, and S. Roukos, Performance of the IBM large vocabulary continuous speech recognition system on the ARPA wall street journal task. In Proc. of ICASSP ’95, pages 41–44, Detroit, MI, 1995. [2] S. Agarwal, S. Godbole, D. Punjani and S. Roy, How Much Noise is too Much: A Study in Automatic Text Classification', In Proc. of ICDM 2007

Has Missing Values?

No

Reviews

There are no reviews for this dataset yet.

Login to Write a Review
Download
0 citations
1818 views

Creators

Shantanu Godbole

Shourya Roy

License

By using the UCI Machine Learning Repository, you acknowledge and accept the cookies and privacy practices used by the UCI Machine Learning Repository.

Read Policy