Method for detecting Ethereum Pincare cheating contract account fused with multi-modal data features

The invention belongs to the technical field of user center system safety supervision, and particularly relates to an Ethereum Pincare cheating contract account detection method fusing multi-modal data features. Comprising the following steps: S100, collecting byte codes, source codes and transaction data related to a contract account; s200, disassembling the byte code into an operation code, and extracting the characteristics of the operation code; extracting text features from the source code; transaction features are extracted from transaction data generated after contract deployment; s300, performing feature selection on the extracted operation code features to obtain dimension-reduced operation code features; s400, mixing the operation code features, the text features and the transaction features after dimension reduction, and training the mixed features; and S500, performing classification prediction by using the trained model. The method can reduce the feature dimensions, improve the model accuracy, an