paolaos/bisect

Data and code for BiSECT project.

BiSECT

This repository contains the code and resources from our paper.

Repo Structure:

1. bisect: BiSECT dataset of complex-simple pairs.

2. metrics: Automatic evaluation metrics for the task that we used in our work.

3. our_model: Code for our hybrid splitter model. Further details to train or generate output from the pretrained models are available in a README file in the folder.

4. outputs: System outputs.

5. experiments: code for running human evaluation as well as MTurk output files

Each folder contains a README with further details.

For the BERT-intialized Transformer baseline, you can refer to this repo. All the pretrained models are available here.

Citation

Please cite if you use the above resources for your research

@inproceedings{bisect2021,
  title={BiSECT: Learning to Split and Rephrase Sentences with Bitexts},
  author={Kim, Joongwon and Maddela, Mounica and Kriz, Reno and Xu, Wei and Callison-Burch, Chris},
  booktitle={Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP)},
  year={2021}
}