This repository is a research prototype for early detecting fake news shared in social media when evidence or metadata is not present.
MIT License
More details are given in our paper.
scripts
folder has examples for reproducing the study:
scripts/elasticsearch_ops.sh
is used for extracting relevant past fake newsscripts/links_ops.sh
is used for extracting link information on the postsscripts/train_server.sh
is used for training the components of the ECOLTrained models and processed data can be downloaded from this link.
The list of additional datasets for training/testing some of the components of the framework: FakeHealth NELA-2019
Please use the following citation if you use our code:
@chapter{baris2021ecol,
title={ECOL: Early Detection of COVID Lies Using Content, Prior Knowledge and Source Information},
author={Ipek Baris and Zeyd Boukhers},
year={2021},
book={Combating Online Hostile Posts in Regional Languages during Emergency Situation },
publisher={Springer International Publishing},
}
Preprint version of the paper is publicly available.