Dataset Card for WITS

Dataset Summary

WITS (Wikipedia for Italian Text Summarization) is a large-scale dataset for abstractive summarization in Italian, built exploiting Wikipedia articles' structure. WITS contains almost 700,000 Wikipedia articles, together with their human-written summaries.

Compared to existing data for text summarization in Italian, WITS is more than an order of magnitude larger and more challenging, given its lengthy sources.

Languages

The dataset is in Italian.

Licensing Information

The dataset uses text from Wikipedia. Please refer to Wikipedia's license.

Citation Information

If you use the dataset, please cite:

@inproceedings{DBLP:conf/clic-it/CasolaL21,
  author={Silvia Casola and Alberto Lavelli},
  title={WITS: Wikipedia for Italian Text Summarization},
  year={2021},
  cdate={1609459200000},
  url={http://ceur-ws.org/Vol-3033/paper65.pdf},
  booktitle={CLiC-it},
  crossref={conf/clic-it/2021}
}