MPS-Bot

The source code of MPS-Bot (Message Passing over Simplexes for Bot Detection) in WWW2026.

Environment Setup

1. Create and activate a conda environment

   conda create --name mpsbot python=3.7
   conda activate mpsbot

2. Install required dependencies:

   wget https://download.pytorch.org/whl/cu102/torch-1.9.1%2Bcu102-cp37-cp37m-linux_x86_64.whl
   pip install torch-1.9.1+cu102-cp37-cp37m-linux_x86_64.whl
   
   wget https://data.pyg.org/whl/torch-1.9.0%2Bcu102/torch_scatter-2.0.9-cp37-cp37m-linux_x86_64.whl
   pip install torch_scatter-2.0.9-cp37-cp37m-linux_x86_64.whl
   
   wget https://data.pyg.org/whl/torch-1.9.0%2Bcu102/torch_sparse-0.6.12-cp37-cp37m-linux_x86_64.whl
   pip install torch_sparse-0.6.12-cp37-cp37m-linux_x86_64.whl
   
   pip install torch-geometric==2.0.2

Dataset Preparation

This project supports two datasets: MGTAB and ​TwiBot22​. Prepare the datasets as follows:

1. MGTAB Dataset

• Download the preprocessed .pt files from the official MGTAB repository.
• Place all files into: Dataset/MGTAB/

2. TwiBot22 Dataset

• Generate dataset files using the code from TwiBot-22/src/BotRGCN.
• After generation, ensure the following files are available: train_idx.pt, test_idx.pt, val_idx.pt, label.pt, edge_index.pt, edge_type.pt, num_properties_tensor.pt, cat_properties_tensor.pt, tweets_tensor.pt, des_tensor.pt
• Place all files into: Dataset/TwiBot22/

Running the Code

Use the following commands to train the model on each dataset:

For MGTAB

   python train.py --dataset mgtab --epochs 200 --lr 1e-3 --dropout 0.1

For TwiBot22

   python train.py --dataset twibot22 --epochs 200 --lr 1e-3 --dropout 0.5

If you use our algorithms in your research, please cite the following papers as reference in your publicaions:

@inproceedings{li2026unmasking,
  title={{U}nmasking {B}ots in {H}igher {D}imensions: {M}essage {P}assing over {S}implexes for {B}ot {D}etection},
  author={Li Fangfang, Zhang Huihui, Zhang Xin and Wu Wei},
  booktitle={WWW},
  pages={4713--4721},
  year={2026}
}