This repository contains the source code of the Neural Reading Comprehension-based Question Answering system, AMANDA, which is published in AAAI 2018.

We experimented on three datasets: NewsQA, TriviaQA and SearchQA. Source code is available in the corresponding directories (NEWSQA, TRIVIAQA, and SEARCHQA) for training and testing on every dataset. Additionally, each of the directories contains a separate README file for further details. amanda/ contains the source code of the model architecture.

If you use the source code or models from this work, please cite our paper:

@article{kundu2018amanda,
  author    = {Kundu, Souvik and Ng, Hwee Tou},
  title     = {A Question-Focused Multi-Factor Attention Network for Question Answering},
  booktitle = {Proceedings of the Thirty-Second {AAAI} Conference on Artificial Intelligence},
  month     = {February},
  year      = {2018},
}

You can also find the paper in arxiv: https://arxiv.org/abs/1801.08290

AMANDA is licensed under the GNU General Public License Version 3. Separate commercial licensing is also available. For more information contact:

• Souvik Kundu ([email protected])
• Hwee Tou Ng ([email protected])