Answers are grounded in this paper and cite their sources.
Accurate detection of security vulnerabilities in smart contracts is crucial to prevent significant financial losses. However, existing deep learning-based methods have not fully exploited the inherent syntactic, semantic, and temporal dependency information in smart contracts, limiting their detection performance. To address this issue, we propose CTVD, a vulnerability detection method that combines the comprehensive code graph (CCG) and the temporal feature enhancement network (TFEN) to enhance detection accuracy. Specifically, the CCG is constructed by integrating abstract syntax trees, control flow graphs, and data flow graphs to construct graphical representations of smart contract functions, and is then optimized by removing redundant information based on vulnerability-related syntactic features. TFEN is subsequently employed to extract temporal features from the CCG, expand feature dimensions, and dynamically assign weights, enabling efficient and automated vulnerability detection. Experimental results show that CTVD excels in detecting reentrancy, timestamp dependency, infinite loop, integer overflow, access control, transaction order dependency, and unchecked external call vulnerabilities, achieving accuracy rates of 92.19%, 91.32%, 89.57%, 91.80%, 81.86%, 80.88%, and 87.11%, respectively, and outperforming existing baseline methods.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: