/SimDial

Synthetic task-oriented dialog generator with controllable complexity. It is released by Tiancheng Zhao (Tony) from Dialog Research Center, LTI, CMU

Primary LanguagePythonApache License 2.0Apache-2.0

CMU SimDial: Synthetic Task-oriented Dialog Generator with Controllable Complexity

This is the dialog data used by our SIGDIAL 2018 paper: Zero-Shot Dialog Generation with Cross-Domain Latent Actions. See paper for details. The source code and data used for the paper can be found at here.

Prerequisites

  • Python 3.6
  • Numpy
  • NLTK
  • progressbar2

Usage

Run the following code to generate dialog data for multiple domains that are defined in the multiple_domains.py script.

python multiple_domains.py

The data will be saved into two folders

test/ for testing data 
train/ for training data

References

If you use any source codes or datasets included in this toolkit in your work, please cite the following paper. The bibtex are listed below:

@article{zhao2018zero,
  title={Zero-Shot Dialog Generation with Cross-Domain Latent Actions},
  author={Zhao, Tiancheng and Eskenazi, Maxine},
  journal={arXiv preprint arXiv:1805.04803},
  year={2018}
}