SmartConDetect: Highly Accurate Smart Contract Code Vulnerability Detection Mechanism using BERT

semanticscholar(2021)

引用 13|浏览9
暂无评分
摘要
Many popular blockchain platforms support smart contracts, which are the programs executed, and stored as transactions on their blockchain protocols and execution environments. However, it is not easy to develop secure smart contracts since smart contracts are programs that can often have security vulnerabilities, which may lead to severe financial loss to service providers or users. Therefore, it is critical to detect security vulnerabilities in smart contracts. In this paper, we propose SmartConDetect to detect security vulnerabilities in smart contracts written in Solidity. SmartConDetect is designed as a static analysis tool to extract code fragments from smart contracts in Solidity and further detect vulnerable code patterns using a pre-trained BERT model. To show the feasibility of SmartConDetect, we evaluate the performance of our approach with 10,000 real-world smart contracts collected from the Ethereum blockchain platform. Our experimental results demonstrate that SmartConDetect outperforms all state-of-the-art methods.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要