Detecting Mixing Services via Mining Bitcoin Transaction Network With Hybrid Motifs

As the first decentralized peer-to-peer (P2P) cryptocurrency system allowing people to trade with pseudonymous addresses, Bitcoin has become increasingly popular in recent years. However, P2P and nature of make transactions on this platform very difficult track, thus triggering emergence various illegal activities ecosystem. Particularly, mixing services Bitcoin, originally designed enhance transaction anonymity, have been widely employed for money laundry complicate trailing illicit fund. In paper, we focus detection addresses belonging services, which is an important task anti-money laundering Bitcoin. Specifically, provide a feature-based network analysis framework identify statistical properties from three levels, namely, level, account level level. To better characterize patterns different types propose concept Attributed Temporal Heterogeneous motifs (ATH motifs). Moreover, deal issue imperfect labeling, tackle as Positive Unlabeled learning (PU learning) problem build model by leveraging considered features. Experiments real datasets demonstrate effectiveness our importance hybrid including ATH detection.