Multilingual Unsupervised Sentence Simplification

Code and pretrained models to reproduce experiments in "MUSS: Multilingual Unsupervised Sentence Simplification by Mining Paraphrases".

Prerequisites

Linux with python 3.6 or above.

Installing

git clone [email protected]:facebookresearch/muss.git
cd muss/
pip install -e .

How to use

Some scripts might still contain a few bugs, if you notice anything wrong, feel free to open an issue or submit a Pull Request.

Simplify sentences from a file using pretrained models

# English
python scripts/simplify.py scripts/examples.en --model-name muss_en_wikilarge_mined
# French
python scripts/simplify.py scripts/examples.fr --model-name muss_fr_mined
# French
python scripts/simplify.py scripts/examples.es --model-name muss_es_mined

Pretrained models should be downloaded automatically, but you can also find them here:
muss_en_wikilarge_mined
muss_en_mined
muss_fr_mined
muss_es_mined

Mine the data

python scripts/mine_sequences.py

Train the models

python scripts/train_model.py

Evaluate simplifications

Please head over to EASSE for Sentence Simplification evaluation.

License

The MUSS license is CC-BY-NC. See the LICENSE file for more details.

Authors

• Louis Martin ([email protected])

Citation

If you use MUSS in your research, please cite MUSS: Multilingual Unsupervised Sentence Simplification by Mining Paraphrases

@article{martin2021muss,
  title={MUSS: Multilingual Unsupervised Sentence Simplification by Mining Paraphrases},
  author={Martin, Louis and Fan, Angela and de la Clergerie, {\'E}ric and Bordes, Antoine and Sagot, Beno{\^\i}t},
  journal={arXiv preprint arXiv:2005.00352},
  year={2021}
}