Robust Fake News Detection using Large Language Models under Adversarial Sentiment Attacks

This repository is the official implementation of AdSent Robust Fake News Detection using Large Language Models under Adversarial Sentiment Attacks published in WWW 2026.

Requirements

To install requirements:

conda create -n Adsent python=3.8.20
conda install pytorch==2.4.1 cudatoolkit=11.3.1 -c pytorch
pip install -r requirements.txt

Dataset

Additional details for Politifact and Gossipcop dataset can be found on their official website.

Evaluation

Credit

This repository is built by Sahar Tahmasebi.

Contributing

Our work is licenced under the CC BY 4.0.