FA-GNN: Filter and Augment Graph Neural Networks for Account Classification in Ethereum

Abstract

As the first blockchain platform supporting smart contracts, Ethereum has become increasingly popular in recent years and generates a massive number of transaction records. Nowadays, millions of accounts in Ethereum have been reported to participate in a variety of businesses, and some of them have been found to be involved in illegal behaviors or even cyber-crimes by exploiting the pseudonymous nature of blockchain. Therefore, there is an urgent need for an effective method to conduct account classification and audit transaction behaviors of each account. In this paper, we model the Ethereum transaction records as a transaction network, and the account classification problem is converted to a node classification problem. Based on the Ethereum transaction network, we propose a novel framework named <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>F</u></b> ilter and <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>A</u></b> ugment <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>G</u></b> raph <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>N</u></b> eural <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>N</u></b> etwork (FA-GNN), which can retain the information of important neighbors and augment node features with high-order information. Experimental results demonstrate that our proposed FA-GNN outperforms state-of-the-art methods in Ethereum account classification.