NFT address clustering and entity tracing method based on heuristic learning

In response to the limitations of the single dimension of entity traceability of Ethereum NFT addresses, this paper proposes a solution that integrates on-chain transaction clustering and off-chain social data traceability. On-chain address clustering is performed based on heuristic rules such as exchange deposit address reuse, token airdrops, and self-authorization behavior, and then addresses are associated with social entities through multi-source heterogeneous data such as social account tag libraries and NFT accounts. The experimental data in this paper is Ethereum transaction data from July 2022 to June 2024. Through clustering analysis, it is concluded that more than 10% of entities control multiple addresses, and a knowledge database associating addresses and entities is further constructed. The research in this paper provides support for subsequent aspects such as regulatory compliance, risk control modeling, and key user tracking in NFT and other similar blockchain-derived scenarios. In the future, anti-clustering identification will be optimized and cross-chain analysis will be conducted to expand the associated knowledge database.