Generative Entity-to-Entity Stance Detection with Knowledge Graph Augmentation

Introduction

This repository contains codes and dataset for paper "Generative Entity-to-Entity Stance Detection with Knowledge Graph Augmentation" (In Proceeding of the 2022 Conference on Empirical Methods in Natural Language Processing).

Set up

Run the following commands to clone the repository.

$ git clone https://github.com/launchnlp/SEESAW.git

Before running our codes, please run the following script to have all dependencies set up.

$ bash requirements.sh

Data

Raw SEESAW can be found under SEESAW directory. Please read README under SEESAW directory for more information.

Processed data and the script for data processing can be found under data directory.

For the data used for Task B: Stance-only prediction for pairwise entities. (see more details in Section 5.1 in our paper), please download from the original data source.

Experiments: Generative Entity-to-Entity Stance Detection

We are still refactoring and cleaning the codes,. Please stay tuned for more updates.

Citation

Please cite our paper if you use our codes and/or SEESAW dataset:

@inproceedings{zhang-etal-2022-seesaw,
    title = "Generative Entity-to-Entity Stance Detection with Knowledge Graph Augmentation",
    author = "Zhang, Xinliang Frederick  and
      Beauchamp, Nicholas  and
      Wang, Lu",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, {EMNLP} 2022",
    year = "2022",
    publisher = "Association for Computational Linguistics",
}

Please also cite the following paper if you run POLITICS as your backbone model:

@inproceedings{liu-etal-2022-politics,
    title = "{POLITICS}: Pretraining with Same-story Article Comparison for Ideology Prediction and Stance Detection",
    author = "Liu, Yujian  and
      Zhang, Xinliang Frederick  and
      Wegsman, David  and
      Beauchamp, Nicholas  and
      Wang, Lu",
    booktitle = "Findings of the Association for Computational Linguistics: {NAACL} 2022",
    year = "2022",
    publisher = "Association for Computational Linguistics",
    pages = "1354--1374",
}

Contact

If you have any question, please contact Xinliang Frederick Zhang <[email protected]> or create a Github issue.