Datasets:

Modalities:
Text
Formats:
csv
Languages:
Urdu
DOI:
Libraries:
Datasets
pandas
License:
XSUMUrdu-DW_BBC / README.md
mbshr's picture
Update README.md
5bc64bb
|
raw
history blame
1.59 kB
metadata
license: cc-by-4.0
task_categories:
  - summarization
language:
  - ur
tags:
  - Urdu
  - Summarization
pretty_name: Urdu Summarization (BBC and DW Urdu News)
size_categories:
  - 10K<n<100K

Urdu_DW-BBC-512

Dataset Description

Dataset Summary

  • Urdu Summarization Dataset containining 76,637 records of Article + Summary pairs scrapped from BBC Urdu and DW Urdu News Websites.
  • Preprocessed Version: upto 512 tokens (~words); removed URLs, Pic Captions etc

Supported Tasks and Leaderboards

Summarization: Extractive and Abstractive

Languages

Urdu.

Data Fields

- url: URL of the article from where it was scrapped (BBC Urdu URLs in english topic text with number & DW Urdu with Urdu topic text)
  dtype: {string}
- Summary: Short Summary of article written by author of article like highlights.
  dtype: {string}
- Text: Complete Text of article which are intelligently trucated to 512 tokens.
  dtype: {string}

Citation Information

https://doi.org/10.48550/arXiv.2310.02790