Research On Ethereum Account Classification Model

Blockchain platform Ethereum has involved millions of accounts due to its strong potential for providing numerous services based on smart contracts.These massive accounts can be divided into diverse categories,such as miners,tokens,and exchanges,which is termed as account diversity.The benefit of investigating diversity are multi-fold,including understand the Ethereum ecosystem deeper and open the possibility of tracking certain abnormal activities.Unfortunately,the exploration of Ethereum account diversity remains scarce.Even the most relevant studies,which focus on the classification of the accounts on Bitcoin,can hardly be applied on Ethereum since their underlying protocols and user idioms are different.To this end,an Ethereum account classification model based on the heuristics and Maximum Likelihood Estimation(MLE)is proposed for the first time.Abnormal features of the Ethereum accounts is summarized systematically from four aspects: transaction frequency,transaction value,transaction time,and transaction data.Furthermore,through in-depth analysis of account behavior in each category and dependency between account categories,a two-stage heuristic algorithm is designed to classify Ethereum accounts.However,the heuristic algorithm fails to take advantage of the accounts' neighbor information and the coverage rate of the heuristic algorithm is relatively low.To solve this problem,when the information of the neighbor nodes is pretty enough,the maximum likelihood estimation principle is used to infer category of the account further.Real-world data through extensive efforts is collected to evaluate the proposed method and show its effectiveness.The validation results show that at most 95.5% of accounts can be classified with accuracy as 84.4%.Based on the proposed account classification model,Ethereum is analyzed to verify the practicality of the account classification model.First,the dynamic evolution of the Ethereum network is analyzed and the following conclusions are made:(1)standalone miners were quickly replaced by mining pools and co-miners,which weakened the security of the Ethereum largely;(2)only about thirty percent standalone miners choose to join the mining pools while most of them choose to leave directly,and the ratio of it does not change over time;(3)transactions related to the mining pools and exchanges take up a large share of the total transactions,which is frustratingly similar to that on Bitcoin.Second,the arbitrage behavior on the Ethereum was uncovered.The results showed that there exists a large number of arbitrage transactions transferring ether from one exchange to another to make a price difference.