Abstract: [Purpose/Significance] Cryptocurrency transaction is globally anonymous with low cost, which is in favor of terrorists and criminals to transfer funds around the world. Effective supervision of cryptocurrency transactions has become a challenge for academia, industry and relevant law enforcement agencies. [Design/Methodology] Relying on transaction data of Bitcoin as a representative cryptocurrency, this study conducts social network analysis to identify abnormal addresses and transaction patterns, and utilizes machine learning algorithms to identify Bitcoin owners. [Findings/Conclusion] Through the analysis of real cases of Bitcoin terrorist financing, it is proved that our proposed method is effective in identifying terrorist financing entities and detecting typical fund transaction patterns of terrorist organizations in cryptocurrency transactions. [Originality/Value] This paper studies the dark side of cryptocurrency transactions through data mining and analysis, which has both practical significance and theoretical value for the regulation of cryptocurrencies and blockchain.

Key words: Bitcoin, Terrorist financing, Blockchain, Network analysis, Machine learning