Analysis of User Behaviors in Blockchain : A Network Science Perspective

Public blockchains offer transparent transaction records, yet identifying actors and interpreting their activities remains challenging. This presentation brings together my research at three levels: (i) micro-level, using network motifs and graphlets to classify DeFi transaction methods; (ii) meso-level, combining money flow analysis and graph representation learning to identify pseudonymous entities; and (iii) macro-level, using multilayer token networks to analyze cross-token trading strategies. Applications include inferring account roles from motifs, identifying entities from money flow representations, and tracking Alameda Research’s trading behavior before bankruptcy. Together, these approaches translate blockchain data into interpretable insights into actors, operations, and financial strategies.