A new paper on arXiv explores whether the structure of decentralized finance (DeFi) transactions can reveal what kind of operation is being performed. DeFi transactions typically involve multiple interacting components—accounts, smart contracts, and tokens—making them difficult to classify by simple rules. The authors propose using graphlets, small subgraph patterns, to capture the local topology of the transaction network.
The study asks a direct question: can network topology alone identify transaction methods? If successful, this approach could offer a way to label or audit DeFi activity without relying on semantic or textual data. The abstract, however, does not report specific results or conclusions, so the effectiveness of the method remains to be seen.
As with many preprint announcements, the full paper would be needed to evaluate the methodology and findings. For now, the work highlights a growing interest in applying network science to understand the structure of blockchain-based financial systems.