STS-NN

The implementation of our paper "Deep Spatial-Temporal Structure Learning for Rumor Detection on Twitter" published on Neural Computing and Applications

Requirements

python 3.6.6
numpy==1.17.2
scipy==1.3.1
pytorch==1.1.0
scikit-learn==0.21.3

How to use

Dataset

The main directory contains the directories of two Twitter datasets: twitter15, and twitter16. In each directory, there are:

• label.txt: this file provides the labels of source tweets in a format like: 'label: source tweet ID';
• data.strnn.vol5000.txt : this file provides the temporal and spatial information of message propagations in a format like: 'source tweets ID\t index-of-the-current-tweet\t index-of-parent-tweet-list(split by ' ')\t index-of-perior-time-tweet(split by ' ')\t text-length\t list-of-index-word-belong-to-current-tweet'
• data.strnn.vol5000.f*.et(tc)*.txt: these files provide the temporal and spatial information of message propagation in the early situation as the format like data.strnn.vol5000.txt file.

In the nfold directory, there are:

• RNNtrainSet_twitter15(6)_tree.txt, RNNtestSet_twitter15(6)_tree.txt : these files provide the training and testing source tweet ID.

These datasets are preprocessed according to our requirement and original datasets can be available at here

Training & Testing

sh run.sh 0 twitter15[twitter16]

The Training & Testing of Early Detection Task

sh run.sh 0 twitter15(6) '' '' '' '' 0(60,120,240,480,720,1440,2160)
sh run.sh 0 twitter15(6) '' '' '' '' '' 0(10,20,40,60,80,200,300)

Citation

If you find the code is useful for your research, please cite this paper:

@article{huang2020deep,
  title={Deep spatial--temporal structure learning for rumor detection on Twitter},
  author={Huang, Qi and Zhou, Chuan and Wu, Jia and Liu, Luchen and Wang, Bin},
  journal={Neural Computing and Applications},
  pages={1--11},
  year={2020},
  publisher={Springer}
}