Official implementation of the Fake News Revealer paper
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Faeze 9ebc0db415 bug in evaluation fixed 1 month ago
data bug in evaluation fixed 1 month ago
.gitignore some changes to store losses and embeddings 1 month ago
LICENSE Initial commit 1 year ago
README.md readme updated 10 months ago
data_loaders.py lime added 8 months ago
evaluation.py bug in evaluation fixed 1 month ago
image.py codes added 10 months ago
learner.py some changes to store losses and embeddings 1 month ago
lime_main.py lime added 8 months ago
main.py fixing a error in configs and main 1 month ago
model.py storing different losses added 1 month ago
optuna_main.py lime added 8 months ago
requirements.txt codes added 10 months ago
test_main.py bug in evaluation fixed 1 month ago
text.py codes added 10 months ago
torch_main.py lime added 8 months ago
utils.py lime added 8 months ago

README.md

Fake News Revealer

Official implementation of the “Fake News Revealer (FNR): A Similarity and Transformer-Based Approach to Detect Multi-Modal Fake News in Social Media” paper (ArXiv Link).

Requirments

FNR is built in Python 3.6 using PyTorch 1.8. Please use the following command to install the requirements:

pip install -r requirements.txt

How to Run

First, place the data address and configuration into the config file in the data directory, and then follow the train and test commands.

train

To run with Optuna for parameter tuning use this command:

python main --data "DATA NAME" --use_optuna "NUMBER OF OPTUNA TRIALS" --batch "BATCH SIZE" --epoch "EPOCHS NUMBER"

To run without parameter tuning, adjust your parameters in the config file and then use the below command:

python main --data "DATA NAME" --batch "BATCH SIZE" --epoch "EPOCHS NUMBER"

test

In the test step, at first, make sure to have the requested ‘checkpoint’ file then run the following line:

python main --data "DATA NAME" --just_test "REQUESTED TRIAL NUMBER"

Bibtex

Cite our paper using the following bibtex item:

@misc{ghorbanpour2021fnr,
      title={FNR: A Similarity and Transformer-Based Approach to Detect Multi-Modal Fake News in Social Media}, 
      author={Faeze Ghorbanpour and Maryam Ramezani and Mohammad A. Fazli and Hamid R. Rabiee},
      year={2021},
      eprint={2112.01131},
      archivePrefix={arXiv},
}