Fraud Detection over Ethereum Blockchain

Abstract

As early as 2013, Vitalik Buterin introduced Ethereum with the possibility of its widespread use (Antonopoulos, 2018). There are several applications based on the Ethereum protocol, including ERC-20 tokens, which are Ethereum-based tokens that can be created and deployed on the Ethereum network. Over 400 million transactions have been made on Ethereum since its inception. There have been a number of illegal activities related to Ethereum, including smart-Ponzi schemes, phishing, money laundering, and fraud. Detecting and predicting such attacks over blockchain can be achieved through anomaly detection for blockchain. The paper makes a number of contributions; first of all, it proposes a Random Forest Classifier, which is an effective method for detecting illicit accounts on the Ethereum network based on the testing of 8 models of four distinct types (Decision Tree, Random Forest, Gradient Boosting, and Extreme Gradient Boosting); secondly, it gives a multiple linear regression model for estimating total Ethereum transfers; and thirdly, it provides coherent and graphical representations of historical data. The software tool, KNIME is used to execute the statistical tasks. This tool uses visual nodes to do descriptive, predictive and prescriptive analytics (Berthold et al., 2009).