@inproceedings{hownottoevaluateyourdialoguesysteman,
  title = {How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation},
  author = {Chia-Wei Liu and Ryan Lowe and Iulian V. Serban and Michael Noseworthy and Laurent Charlin and Joelle Pineau},
  year = {2016},
  booktitle = {EMNLP 2016 11},
  eprint = {1603.08023},
  archivePrefix = {arXiv},
  url = {http://arxiv.org/abs/1603.08023v2},
}