Research on Smart Contract Vulnerability Detection Technology Based on VCS and Ensemble Learning

Smart contracts play a crucial role in blockchain technology, but their writing poses risks of vulnerabilities, potentially leading to serious consequences such as financial losses and system crashes. To address this, we propose a smart contract vulnerability detection method based on VCS and ensemble learning. This method first utilizes Vulnerability Candidate Slicing (VCS) technology to extract syntax and semantic features, enhancing detection capabilities. Then, it employs Word2vec, FastText, GloVe, and other embedding models to transform raw inputs into vector representations, capturing more semantic information. Finally, an ensemble learning strategy integrates multiple neural network models to improve detection performance and mitigate the limitations of individual models. Experimental results demonstrate that this method outperforms other advanced tools in the market, providing robust support for ensuring the security and stability of blockchain systems.