그래프 분류를 통한 이더리움 피싱 계좌 탐지 기법

Abstract

Ethereum is being exploited for cybercrimes due to the anonymity of blockchain. The phishing that exploits Ethereum steals users’ assets to destabilize the Ethereum ecosystem. Therefore, detecting phishing accounts is important to protect users’ assets. In this paper, we propose a graph neural network-based phishing account detection via graph classification. We conducted experiments to detect phishing accounts on real Ethereum transactions. The experimental results show that the proposed method outperforms previous phishing detection methods by up to 0.12 f1-score.