metadata

license: mit
task_categories:
  - text-classification
  - zero-shot-classification
  - feature-extraction
  - text-generation
language:
  - en
pretty_name: 'CHEW: A Dataset of CHanging Events in Wikipedia'
size_categories:
  - 1K<n<10K

Dataset Card for Dataset Name

This is the lexical/no-overlapping split of the CHEW dataset(CHEW: A Dataset of CHanging Events in Wikipedia).

Dataset Details

Dataset Description

This dataset is the Lexical/No-overlapping split of the CHEW Dataset,where CHEW stands for CHanging Events in Wikipedia. It contains Wikipedia titles, text in two timestamped versions and Binary Label showing Change(1) or No change(0). Change here means there has been informationm change in second version of the text with respect to the first version. The dataset is created for Information change detection of two versions of a Wikipedia article.

The fields in the dataset are:

Title: Wikipedia article title
Text: Contains the two versions of the text separated by comma(','). The format of this column is:
```
<t> TITLE </t> <y> TIMESTAMP1 </y> TEXT1 , <t> TITLE </t> <y> TIMESTAMP2 </y> TEXT2
```
Label: Binary Label containing
- 0: No change/syntactic changes
- 1: Information change
Curated by: Hsuvas Borkakoty
Language(s) (NLP): English
License: MIT

Dataset Sources [optional]

Repository: https://github.com/hsuvas/temporal_wikipedia
Paper [optional]: CHEW: A Dataset of CHanging Events in Wikipedia (Borkakoty and Espinosa-Anke,2024)