Portrait of decentralized application users: an overview based on large-scale Ethereum data

Decentralized application (DApp) is an emerging technology designed to address distrust, privacy, and security issues. However, we note that the research community in human factors has not conducted in-depth research on user behavior in this unique distributed environment yet. In this paper, unlike a small sample of user interviews, we attempt to profile DApp users through publicly available data. Using Ethereum as an example, we build a series of datasets containing more than 73.8 million transactions generated by 230,000 addresses. By transforming hexadecimal addresses into readable application names, we analyze the behavioral characteristics of the user based on the categories of DApp. Furthermore, we apply an unsupervised clustering method on the 230,000 addresses to distinguish investors and players and analyze their behavioral patterns and sensitivity to blockchain markets, such as ETH prices. In addition, we implement heuristics to demonstrate how blockchain data mining can facilitate practical systems, including anomaly detection and recommend systems. Finally, we discuss future directions for studying human factors in a decentralized context and hope that this work will attract more research attention and support the development of DApp and further Metaverse ecosystems.