An Attention-based Wide and Deep Neural Network for Reentrancy Vulnerability Detection in Smart Contracts

Samuel Banning Osei, Rubing Huang, Zhongchen Ma

Abstract: Highlights•Optimize smart contract code to remove extraneous parts and extract key fragments.•Convert optimized snippets into numerical vectors representing semantic features.•Use CBOW to generate dense vector embeddings capturing contextual information.•Train model with vector embeddings to detect reentrancy vulnerabilities.•Leverage an attention mechanism to balance depth and precision in model.