Novel blockchain transaction provenance model with graph attention mechanism

Abstract

With the maturity of the blockchain technology, more and more blockchain digital currencies including the Bitcoin, the Ethereum and the Ripple have been developed. Meanwhile, the security problem of the blockchain digital currency has become increasingly serious. It is necessary to build a provenance model for a large number of blockchain transaction, which can be used to find out which link has trouble and who is responsible once a problem occurs. However, the existing blockchain transaction data analysis methods have low traceback accuracy for data provenance. Therefore, a novel blockchain transaction provenance model with graph attention mechanism is proposed in this paper. The graph attention mechanism is used to identify the traders. Then, a triplet data structure is built to record provenance information and express the relationship between transaction amount and traders. Finally, the smart contract is used to implement the proposed provenance model. The provenance information can be searched through the interface provided by the smart contract when the traceback is required. In addition, the real large-scale Bitcoin transaction dataset Elliptic is used to compare the proposed method with other existing methods. The experiment show that the proposed method achieves state of the art results of the traceback accuracy.