A Semantic Analysis-Based Method for Smart Contract Vulnerability

Abstract

Smart contracts are distributed, self-executing programs on blockchain network and they have the potential to impact many industries. A large number of developers are optimistic about the development prospects of smart contracts. However, due to different levels of familiarity of developers with technology, the existence of serious vulnerabilities in code can result in huge property losses. The serious consequences of reentrancy vulnerabilities have attracted widespread attention from researchers, and researchers have proposed many approaches to detect them. Many experiments show that these approaches can only effectively detect certain types of vulnerabilities. Similar vulnerabilities still exist in some smart contracts. In this paper, we propose a semantic analysis-based detection method for reentrancy vulnerabilities. The method utilizes machine learning techniques to efficiently detect the vulnerabilities of the given code. And then, our method provides semantic analysis-based effective correction feedback of the code. Compared with the existing vulnerability detection tools Slither and Mythril, our method improves the work efficiency. Moreover, compared with the current Mythril method with the best vulnerability detection accuracy, our method improves the accuracy by 1.01%.