An Automated Bitcoin Address Labeling Method Based on Transaction Behavior Analysis

Abstract

Bitcoin employs an anonymity mechanism to protect users' real identities from being exposed, making it widely used in illegal activities such as money laundering, dark web black market transactions, and more. As a result, tracing the source of illegal Bitcoin transactions has become a critical task. A comprehensive Bitcoin address label library is essential for achieving this traceability. However, existing label libraries suffer from low labeling rates and poor label quality. Additionally, current labeling methods are unable to enhance label quality while minimizing labeling costs. To address these issues, we propose an automatic labeling method for Bitcoin transaction addresses based on transaction behavior analysis. Firstly, we analyzed the relationships between Bitcoin miners, mining pools, and the operational modes of these pools. By examining on-chain transaction behavior data, we successfully labeled the addresses of miners and mining pools. Simultaneously, we gathered transaction behavior information from off-chain news media and applied entity recognition and relationship extraction algorithms for automatic address labeling. Experimental results demonstrate that this method can accurately and automatically label Bitcoin addresses, offering advantages such as high accuracy, low labeling cost, and excellent real-time performance.