Transaction Tracking on Blockchain Trading Systems using Personalized PageRank
share
Due to the pseudonymous nature of blockchain, various cryptocurrency systems like Bitcoin and Ethereum have become a hotbed for illegal transaction activities. The ill-gotten profits on many blockchain trading systems are usually laundered into concealed and "clean" fund before being cashed out. Recently, in order to recover the stolen fund of users and reveal the real-world entities behind the transactions, much effort has been devoted to tracking the flow of funds involved in illegal transactions. However, current approaches are facing difficulty in estimating the cost and quantifying the effectiveness of transaction tracking. This paper models the transaction records on blockchain as a transaction network, tackle the transaction tracking task as graph searching the transaction network and proposes a general transaction tracking model named as Push-Pop model. Using the three kinds of heuristic designs, namely, tracking tendency, weight pollution, and temporal reasoning, we rewrite the local push procedure of personalized PageRank for the proposed method and name this new ranking method as Transaction Tracking Rank (TTR) which is proved to have a constant computational cost. The proposed TTR algorithm is employed in the Push-Pop model for efficient transaction tracking. Finally, we define a series of metrics to evaluate the effectiveness of the transaction tracking model. Theoretical and experimental results on realist Ethereum cases show that our method can track the fund flow from the source node more effectively than baseline methods.
READ FULL TEXT
