본문 바로가기
  • Home

Static Detection of Security Vulnerabilities Using Machine Learning on Graph Information

  • Journal of Software Assessment and Valuation
  • Abbr : JSAV
  • 2022, 18(2), pp.1-9
  • DOI : 10.29056/jsav.2022.12.01
  • Publisher : Korea Software Assessment and Valuation Society
  • Research Area : Engineering > Computer Science
  • Received : November 3, 2022
  • Accepted : December 20, 2022
  • Published : December 31, 2022

Ji Hwan Mo 1 Sung-Moon Hong 1 Doh, Kyung-Goo 2

1한양대학교(ERICA캠퍼스)
2한양대학교

Accredited

ABSTRACT

Taint analysis is widely used in detecting security vulnerabilities in source code. However, it is difficult to obtain accurate results in reasonable amount of computing time due to the nature of static analysis. This paper proposes a new way of detecting security vulnerabilities using machine-learning technology utilizing an already established model, graph2vec. The flow graph of a program is transformed into a vector and then given to the model. The data set is prepared by first constructing simplified program and then applying mutation. The evaluation results show that the model detects security vulnerabilities with the accuracy of up to 99%, positively showing the possibility of applying machine-learning technology to the static detection of security vulnerabilities in source code.

Citation status

* References for papers published after 2023 are currently being built.