For NVD Dataset

• Python3, Java runtime 8, sbt (Scala build tool)
• Joern (documents)
• cpgclientlib library for using cpg ($ pip install cpgclientlib)

For SARD Dataset

• LLVM-Slicing

For Evaluation

• HTMLTestRunner for making test reports ($ pip install HTMLTestRunner)
• coverage for checking the test coverage ($ pip install coverage)

• Dataset: Original source code of dataset, snippet files, tokenizing file
• Resource: Slicing criterion file
• src: Main source code of AutoVAS. The src direction has model and preprocessing folder.
• tool: Utility files for AutoVAS such as joern, llvm-slicing

Jeon, S., & Kim, H. K. (2021). AutoVAS: An Automated Vulnerability Analysis System with a Deep Learning Approach. Computers & Security, 102308.

@article{jeon2021autovas,
  title={AutoVAS: An Automated Vulnerability Analysis System with a Deep Learning Approach},
  author={Jeon, Sanghoon and Kim, Huy Kang},
  journal={Computers & Security},
  pages={102308},
  year={2021},
  publisher={Elsevier}
}

The uploaded snippet, which consists of the C language-based snippet, is part of a total snippet. In the NVD dataset, we applied some heuristic points as a slicing criterion such as arithmetic, array, etc., in addition to vulnerable APIs. Lastly, we only uploaded snippets after preprocessing without the program slicing module.

This program is authored and maintained by Sanghoon(Kevin) Jeon.

Email: kppw99@gmail.com

GitHub@kppw99