Skip to content

Code for the experiments in the paper EACL 2021: From the Stage to the Audience: Propaganda on Reddit

License

Notifications You must be signed in to change notification settings

nyxpho/propaganda_eacl2021

Repository files navigation

EACL 2021: From the Stage to the Audience: Propaganda on Reddit

The data required for running this code can be found here: https://propaganda.qcri.org/

This repository contains the following files:

  1. bert-propaganda.py will train a model for classifying sentences in propaganda or non propaganda
  2. bert-technique.py will train a model classifying a propaganda sentences according to the type of propaganda
  3. mediafact.txt contains the labels of newspapers crawled from the site https://mediabiasfactcheck.com/ with their corresponding label
  4. stattertext_topics.py will give the confounding tokens

In order to obtain the Roberta and XLNet models from the paper, you need just to modify the parameter 'version_model' in the python files above. In order to get the MGN ReLU model, please use the same link as for the data.

About

Code for the experiments in the paper EACL 2021: From the Stage to the Audience: Propaganda on Reddit

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages