https://github.com/pubmedqa/pubmedqa

PubMedQA: A Dataset for Biomedical Research Question Answering
https://github.com/pubmedqa/pubmedqa

Last synced: 3 months ago
JSON representation

PubMedQA: A Dataset for Biomedical Research Question Answering

• Host: GitHub
• URL: https://github.com/pubmedqa/pubmedqa
• Owner: pubmedqa
• License: mit
• Created: 2019-08-23T12:48:40.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2023-04-18T13:19:13.000Z (over 1 year ago)
• Last Synced: 2024-06-11T05:34:25.059Z (5 months ago)
• Language: Python
• Homepage: https://pubmedqa.github.io
• Size: 688 KB
• Stars: 216
• Watchers: 5
• Forks: 27
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• StarryDivineSky - pubmedqa/pubmedqa

README

        
# PubMedQA

## Download

PQA-L is already in `./data/`

[PQA-U](https://drive.google.com/open?id=1RsGLINVce-0GsDkCLDuLZmoLuzfmoCuQ)

[PQA-A](https://drive.google.com/open?id=15v1x6aQDlZymaHGP7cZJZZYFfeJt2NdS)

## Split the dataset

After downloading PQA-A and PQA-U as `ori_pqaa.json` and `ori_pqau.json` in the `./data/`, enter the `./preprocess/` directory and split the dataset:

```bash

cd preprocess

python split_dataset.py pqaa

python split_dataset.py pqal

```

Please be aware that there is no offical code for splitting PQA-U.

## Evaluation and submission

To evaluate your model predictions, please prepare the results in a json format where the key is PMID and value is one of "yes", "no", and "maybe". Run the following script to get the performance:

```bash

python evaluation.py PREDICTIONS_PATH

```

To submit a system on the Leaderboard, please send an email that contains the model predictions and a brief description of the system to Qiao Jin via [[email protected]](mailto:[email protected]).

## Human performance

After splitting the PQA-L and having `./data/test_set.json`, one can run the following script to get human performance:

```bash

python get_human_performance.py

```

## Citation

If you use PubMedQA in your research, please cite our paper by:

```

@inproceedings{jin2019pubmedqa,

  title={PubMedQA: A Dataset for Biomedical Research Question Answering},

  author={Jin, Qiao and Dhingra, Bhuwan and Liu, Zhengping and Cohen, William and Lu, Xinghua},

  booktitle={Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)},

  pages={2567--2577},

  year={2019}

}

```