Enhancing Change Point Detection Through Key Actor Identification: A Study in Cryptocurrency Transactions Networks

This paper proposes a unified framework for analyzing cryptocurrency transaction networks, employing a composite approach that combines key actor identification and change point detection. At first, the framework identifies a representative key actor in a network of cryptocurrency transactions and retrieves its relevant transactions. Then, multiple features are extracted in the form of time series from its transaction history, providing an overview of its activity over time. In addition, feature selection is applied to reduce overlapping information among the extracted features and, finally, change point detection is utilized to identify time locations of significant changes in the selected features. To showcase the applicability of the proposed framework, four cryptocurrency transaction networks are used related to the tokens DAI, WLUNA, USTC and PAX. The analysis results indicate that the estimated change points often coincide with market events that may have influenced the network’s transaction behavior. Overall, this paper presents a new approach for understanding the dynamics of cryptocurrency transaction networks, where, by identifying and examining their representative key actors, we aim to obtain a comprehensive characterization of its overall structure.