【Author】
Sun, Hanyi; Ruan, Na; Liu, Hanqing
【Source】NETWORK AND SYSTEM SECURITY, NSS 2019
【Abstract】As an open source public blockchain with the capabilities of running smart contract, Ethereum provides decentralized Ethernet virtual machines to handle peer-to-peer contracts through its dedicated cryptocurrency Ether. And as the second largest blockchain, the amount of transaction data in Ethereum grows fast. Analysis of these data can help researchers better understand Ethereum and find attackers among the users. However, the analysis of Ethereum data at the present stage is mostly based on the statistical characteristics of Ethereum nodes and lacks analysis of the transaction behavior between them. In this paper, we apply machine learning in Ethereum analysis for the first time and cluster users and smart contract into groups by using transaction information in existing blocks. The clustering results are analyzed by using the identity information of the available Ethereum users and smart contracts. Based on the clustering results, we propose a new way of user identity discrimination and malicious user detection.
【Keywords】Blockchain; Ethereum; Network embedding
【摘要】作为一个具有运行智能合约能力的开源公共区块链，以太坊提供去中心化的以太网虚拟机，通过其专用加密货币以太坊处理点对点合约。作为第二大区块链，以太坊的交易数据量增长迅速。分析这些数据可以帮助研究人员更好地了解以太坊，并在用户中找到攻击者。然而，现阶段对以太坊数据的分析大多基于以太坊节点的统计特征，缺乏对节点之间交易行为的分析。在本文中，我们首次将机器学习应用于以太坊分析，利用现有区块中的交易信息将用户和智能合约集群到组中。聚类结果通过使用可用的以太坊用户和智能合约的身份信息进行分析。基于聚类结果，提出了一种新的用户身份识别和恶意用户检测方法。
评论