【摘要】 随着区块链技术的发展,智能合约的作用愈加重要,它通过降低合作中的信任成本,显著提升了系统的可信度,创造了数以亿计的经济价值。然而,这一巨大价值也吸引了大量的攻击事件,伴随攻击频率的增加,智能合约安全检测工具的重要性愈发凸显。本文首先详细介绍区块链技术及智能合约的基本概念;其次,从代码层面、虚拟机层面和区块链层面深入分析智能合约漏洞;然后,讨论传统漏洞检测方法的实现原理和存在的缺陷,以及基于机器学习和深度学习的检测技术存在的数据集不平衡和泛化能力差等问题;最后,本文总结所讨论的内容,并根据存在的问题展望未来的发展方向。更多还原

【Abstract】 With the development of blockchain technology, the role of smart contracts has become increasingly important. They significantly enhance the credibility of systems by reducing trust costs in collaborations, thereby creating substantial economic value. However, this immense value has also attracted a plethora of attack incidents. As the frequency of these attacks rises, the importance of smart contract security detection tools becomes even more pronounced. This article first provides a detailed introduction to the fundamental concepts of blockchain technology and smart contracts. Secondly, it conducts an in-depth analysis of smart contract vulnerabilities from the perspectives of code, virtual machines and blockchain layers. It then discusses the principles of traditional vulnerability detection methods, their inherent flaws, and the issues associated with machine learning and deep learning-based detection technologies such as dataset imbalance and poor generalization capabilities. Finally, the article summarizes the discussed content and looks forward to future development directions based on the existing problems.更多还原