Early-stage phishing detection on the Ethereum transaction network

As cryptocurrency is widely accepted and used, attendant illegal activities have attracted extensive attention, especially phishing scams, which bring great losses to both customers countries. From the perspective of crime prevention, early warning such behaviors significance. However, most existing studies focus on detecting scams that already occurred been reported. In addition, previous ignore temporal order users' appearance thus cannot accurately extract features reflecting users’ transaction patterns. this paper, we propose a framework called early-stage detection address problem detection. According amount, first divide process into three stages: stage, middle late stage. Then, develop feature extraction method capture from local network structures time series transactions. experiments, dataset strictly partitioned by series, experimental results show our proposed outperforms graph embedding methods real-world Ethereum dataset. Finally, select ten important analyze differences between users normal these features, provide useful insights for regulators platforms detect in advance.