Threat Prediction of Abnormal Transaction Behavior Based on Graph Convolutional Network in Blockchain Digital Currency

There are some malicious traders in the current blockchain digital currency market, and their abnormal transaction behaviors have seriously threatened the security of a large number of users’ assets. Therefore, the research on the threat prediction method of abnormal transaction behavior of blockchain digital currency is beneficial to actively prevent the threat of abnormal transaction behavior, and is of great significance to maintain the ecological environment health of blockchain digital currency. The representative Bitcoin is a currency with the highest market value among many blockchain digital currencies, with a huge number of users and transactions. The Bitcoin trading system is dynamic, and different types of abnormal transaction behaviors have different structures, so it is extremely challenging to predict the threat of abnormal transaction behaviors. This paper designs a new threat prediction method for abnormal transaction behavior. Specifically, through the analysis of the transaction relation pattern of typical abnormal transaction behaviors of Bitcoin, the abnormal transaction behaviors are uniformly modeled as an object-relation pattern. Then, the research proposes TSRGL framework, which uses R-GCN to learn the topology structure of the historical object-relation snapshot graph, and realizes the threat prediction of abnormal transaction behavior. To evaluate the effectiveness of this approach, we verified the effectiveness of the TSRGL framework based on real Bitcoin abnormal transaction behavior dataset. The experimental results show that TSRGL framework has better performance than the baseline methods and is currently the most suitable framework for the threat prediction of abnormal transaction behavior.