mlsum_de_train_10k / README.md
giuliadc's picture
Update README.md
9c2b182 verified
metadata
task_categories:
  - summarization
language:
  - de
size_categories:
  - 1K<n<10K

Train split of german MLSUM filtered by using the code by Aumiller et al. (1) available at https://github.com/dennlinger/summaries/tree/main with following settings:

min_length_summary=15, min_length_reference=200, length_metric="whitespace", extractiveness="fully"

After this filtering operation, the dataset was filtered again to exclude all samples with bigram overlap fraction between summary and article >= 0,76.

Finally, 10k random saples were selected and kept in this dataset. All other samples were removed.

(1): Aumiller, D., Fan, J., & Gertz, M. (2023). On the State of German (Abstractive) Text Summarization. arXiv preprint arXiv:2301.07095.