Defining user spectra to classify Ethereum users based on their behavior

Abstract Purpose In this paper, we define the concept of user spectrum and adopt it to classify Ethereum users based on their behavior. Design/methodology/approach Given a time period, our approach associates each with showing trend some behavioral features obtained from social network-based representation Ethereum. Each class has its own spectrum, by averaging spectra users. order evaluate similarity between one user, propose tailored measure adapting context general measures provided in past. Finally, test dataset transactions. Findings We model represent also for (i.e., token contract, exchange, bancor uniswap), consisting suitable multivariate series. Furthermore, an new The core is metric capable measuring degree This Eros distance Extended Frobenius Norm) scenario. Originality/value paper introduces users, which blockchains. Differently past models, represented behavior means univariate series, here proposed exploits Moreover, shows that original does not return satisfactory results when applied spectra, proposes modified version it, reference scenario, reaches very high accuracy. adopts Currently, no multi-class automatic classification exists yet, albeit single-class ones have been recently proposed. Therefore, only way are online services (e.g., Etherscan), where classified after request them. However, fraction thus low. To address issue, present