Smart contract vulnerability detection based on semantic graph and residual graph convolutional networks with edge attention
- Unzip
/utils/python-solidity-parser-master.zip, and runpython3 setup.py install. - Configure the source code path in
/utils/Sourcecode2AST.ipynbbefore running it. - Run
SG.ipynbto convert AST files to Semantic Graph and Edge Series. - Run
BFS_EA_RGCN(SG).ipynbto train the model to get the final result.
- scikit-learn==1.1.1
- scipy==1.5.0
- torch-geometric==2.2.0
- torch==1.9.1
- tqdm==4.47.0
- /utils/layer.py
- /utils/utils.py
- /utils/pytorchtools.py
If you find this work helpful, please kindly cite our paper.
@article{CHEN2023111705,
title = {Smart contract vulnerability detection based on semantic graph and residual graph convolutional networks with edge attention},
journal = {Journal of Systems and Software},
volume = {202},
pages = {111705},
year = {2023},
issn = {0164-1212},
doi = {https://doi.org/10.1016/j.jss.2023.111705},
url = {https://www.sciencedirect.com/science/article/pii/S0164121223001000},
author = {Da Chen and Lin Feng and Yuqi Fan and Siyuan Shang and Zhenchun Wei},
keywords = {Smart contract vulnerability detection, Code graph, Graph convolutional networks, Edge attention, Residual block}
}