/DM-INTER

Source code of our CIKM'24 paper "Why Misinformation is Created? Detecting them by Integrating Intent Features"

Primary LanguagePython

DM-INTER

This repo is the released code of our work Why Misinformation is Created? Detecting them by Integrating Intent Features, published in CIKM'24.

🎉 [ 2024/07 ] This paper is accepted by CIKM 2024.

Our released code follows to "Generalizing to the Future: Mitigating Entity Bias in Fake News Detection".

Requirements

torch==1.12.1
cudatoolkit==11.3.1
transformers==4.27.4

Prepare Datasets

You can download the dataset GossipCop from ENDEF, SIGIR 2023, and PolitiFact and Snopes from https://www.mpi-inf.mpg.de/dl-cred-analysis/ , and then place them to the folder ./data;

Train

  • Run the shell script:
python main.py --model_name t5our --dataset gossip

where --dataset includes gossip, politifact, snopes; --model_name contains t5, t5emo, t5mdfend, t5our (t5ours represents our proposed DM-INTER model).

  • Check log files in ./log, and we prepare an automatic tool read_results.py to convert log files to an excel table.

Citation

  • ArXiv version
@article{wang2024why,
  author       = {Bing Wang and
                  Ximing Li and
                  Changchun Li and
                  Bo Fu and
                  Songwen Pei and
                  Shengsheng Wang},
  title        = {Why Misinformation is Created? Detecting them by Integrating Intent
                  Features},
  journal      = {CoRR},
  volume       = {abs/2407.19196},
  year         = {2024},
}