Code for the ACL 2022 paper: Tackling Fake News Detection by Continually Improving Social Context Representations using Graph Neural Networks

There are two folders, each with a README: "Code" and "Dataset_Release". Code contains the code to build the graph, train it on Node Classification, and run Inference Operators. Dataset_release contains the data that we collected (some of it is hosted on Google Drive and the links are provided).

To cite:

@inproceedings{mehta2022tackling,
  title={Tackling Fake News Detection by Continually Improving Social Context Representations using Graph Neural Networks},
  author={Mehta, Nikhil and Pacheco, Mar{\'\i}a Leonor and Goldwasser, Dan},
  booktitle={Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages={1363--1380},
  year={2022}
}

All data is released as Anonymized. Articles can be provided by emailing [email protected] and agreeing to our terms of use to respect ethical concerns. We do provide a file user_twitter_to_id.tsv where you can map our Graph IDs to twitter IDs so you can download the respective Twitter users if necessary. Sources can be scraped based on the News-Media-Reliability dataset ((https://github.com/ramybaly/News-Media-Reliability)[href]). We will provide scraping scripts soon.