Dataset Card for "mlsum"

Dataset Summary

We present MLSUM, the first large-scale MultiLingual SUMmarization dataset. Obtained from online newspapers, it contains 1.5M+ article/summary pairs in five different languages -- namely, French, German, Spanish, Russian, Turkish. Together with English newspapers from the popular CNN/Daily mail dataset, the collected data form a large scale multilingual dataset which can enable new research directions for the text summarization community. We report cross-lingual comparative analyses based on state-of-the-art systems. These highlight existing biases which motivate the use of a multi-lingual dataset.

Supported Tasks and Leaderboards

More Information Needed

Languages

More Information Needed

Dataset Structure

We show detailed information for up to 5 configurations of the dataset.

Data Instances

de

  • Size of downloaded dataset files: 330.52 MB
  • Size of the generated dataset: 897.34 MB
  • Total amount of disk used: 1227.86 MB

An example of 'validation' looks as follows.

{
    "date": "01/01/2001",
    "summary": "A text",
    "text": "This is a text",
    "title": "A sample",
    "topic": "football",
    "url": "https://www.google.com"
}

es

  • Size of downloaded dataset files: 489.53 MB
  • Size of the generated dataset: 1274.55 MB
  • Total amount of disk used: 1764.09 MB

An example of 'validation' looks as follows.

{
    "date": "01/01/2001",
    "summary": "A text",
    "text": "This is a text",
    "title": "A sample",
    "topic": "football",
    "url": "https://www.google.com"
}

fr

  • Size of downloaded dataset files: 591.27 MB
  • Size of the generated dataset: 1537.36 MB
  • Total amount of disk used: 2128.63 MB

An example of 'validation' looks as follows.

{
    "date": "01/01/2001",
    "summary": "A text",
    "text": "This is a text",
    "title": "A sample",
    "topic": "football",
    "url": "https://www.google.com"
}

ru

  • Size of downloaded dataset files: 101.30 MB
  • Size of the generated dataset: 263.38 MB
  • Total amount of disk used: 364.68 MB

An example of 'train' looks as follows.

{
    "date": "01/01/2001",
    "summary": "A text",
    "text": "This is a text",
    "title": "A sample",
    "topic": "football",
    "url": "https://www.google.com"
}

tu

  • Size of downloaded dataset files: 236.03 MB
  • Size of the generated dataset: 662.79 MB
  • Total amount of disk used: 898.82 MB

An example of 'train' looks as follows.

{
    "date": "01/01/2001",
    "summary": "A text",
    "text": "This is a text",
    "title": "A sample",
    "topic": "football",
    "url": "https://www.google.com"
}

Data Fields

The data fields are the same among all splits.

de

  • text: a string feature.
  • summary: a string feature.
  • topic: a string feature.
  • url: a string feature.
  • title: a string feature.
  • date: a string feature.

es

  • text: a string feature.
  • summary: a string feature.
  • topic: a string feature.
  • url: a string feature.
  • title: a string feature.
  • date: a string feature.

fr

  • text: a string feature.
  • summary: a string feature.
  • topic: a string feature.
  • url: a string feature.
  • title: a string feature.
  • date: a string feature.

ru

  • text: a string feature.
  • summary: a string feature.
  • topic: a string feature.
  • url: a string feature.
  • title: a string feature.
  • date: a string feature.

tu

  • text: a string feature.
  • summary: a string feature.
  • topic: a string feature.
  • url: a string feature.
  • title: a string feature.
  • date: a string feature.

Data Splits

name train validation test
de 220887 11394 10701
es 266367 10358 13920
fr 392902 16059 15828
ru 25556 750 757
tu 249277 11565 12775

Dataset Creation

Curation Rationale

More Information Needed

Source Data

Initial Data Collection and Normalization

More Information Needed

Who are the source language producers?

More Information Needed

Annotations

Annotation process

More Information Needed

Who are the annotators?

More Information Needed

Personal and Sensitive Information

More Information Needed

Considerations for Using the Data

Social Impact of Dataset

More Information Needed

Discussion of Biases

More Information Needed

Other Known Limitations

More Information Needed

Additional Information

Dataset Curators

More Information Needed

Licensing Information

Usage of dataset is restricted to non-commercial research purposes only. Copyright belongs to the original copyright holders. See https://github.com/recitalAI/MLSUM#mlsum

Citation Information

@article{scialom2020mlsum,
  title={MLSUM: The Multilingual Summarization Corpus},
  author={Scialom, Thomas and Dray, Paul-Alexis and Lamprier, Sylvain and Piwowarski, Benjamin and Staiano, Jacopo},
  journal={arXiv preprint arXiv:2004.14900},
  year={2020}
}

Contributions

Thanks to @RachelKer, @albertvillanova, @thomwolf for adding this dataset.

Models trained or fine-tuned on mlsum