Smart Contract Vulnerability Auditor Using GNN

Artificial intelligence and blockchain technology both of these are potent innovations that are revolutionizing several industries right now. While AI allows robots to learn from data and make decisions that are similar to those made by humans, blockchain, with its decentralized record, offers a safe and transparent mechanism to carry out transactions. The integration of these two technologies is the main subject of our research. The security issues related to smart contracts have received a lot of attention due to significant financial losses resulting from vulnerabilities. The accuracy of the current methods for detecting these vulnerabilities is limited because they rely on fixed expert guidelines. In this paper, we suggest smart contract vulnerability identification with graph neural networks (GNNs). To be more precise, we start by building a contract graph that symbolizes the semantic and syntactic structures of a smart contract function. We employ an elimination step to draw attention to the main nodes in order to simplify the network. Next, to learn from the normalized graphs for vulnerability identification, we create a new temporal message propagation network (TMP) and a degree-free graph convolutional neural network (DR-GCN). Comprehensive tests show that our suggested strategy outperforms the state-of-the-art techniques in identifying three distinct kinds of vulnerabilities.