/GTR

[SIGIR 2021] Retrieving Complex Tables with Multi-Granular Graph Representation Learning.

Primary LanguagePythonApache License 2.0Apache-2.0

Graph-based Table Retrieval (GTR)

Code and data for our paper Retrieving Complex Tables with Multi-Granular Graph Representation Learning at SIGIR 2021.

Quick Links

Preliminary

Install DGL:

conda install -c dglteam dgl-cuda10.2  # pay attention to the cuda version

Install fastText:

git clone https://github.com/facebookresearch/fastText.git
cd fastText
pip install .

Download pretrained word vectors:

wget https://dl.fbaipublicfiles.com/fasttext/vectors-wiki/wiki.en.zip
unzip wiki.en.zip

Install trec_eval tool:

git clone https://github.com/usnistgov/trec_eval.git
cd trec_eval
make

Install other requirements:

pip install -r requirements.txt

Run

To run cross validation on WikiTables dataset:

python run.py --exp cross_validation --config configs/wikitables.json

Citation

If you use our code in your research, please cite our work:

@inproceedings{wang2021retrieving,
  title={Retrieving Complex Tables with Multi-Granular Graph Representation Learning},
  author={Wang, Fei and Sun, Kexuan and Chen, Muhao and Pujara, Jay and Szekely, Pedro},
  booktitle={Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval},
  pages={1472--1482},
  year={2021}
}