Analysis And Research Of Ethernet Smart Contract Scam Based On Transaction Characteristics

The increasingly frequent transaction behavior in Blockchain has gradually become a hot topic in the society.On the one hand,the use of virtual electronic currency provides a convenient and safe transaction environment,on the other hand,it also has hidden dangers of anonymity and untraceability.Now with the development of trading platform based on the block chain intensified,generated by the huge trading data and its underlying technology features also provided an opportunity for illegal molecules,for the etheric fang on intelligent contract fraud is difficult to track and predict problems,this paper put forward:1)In view of the illegal account problem in Ethereum,decision tree classification algorithm is used to analyze the transaction data of contract account on Ethereum,extract a variety of attribute characteristics,analyze various data indicators under attribute characteristics,and on this basis,compare the model with other models in the same environment.The results show that the prediction accuracy reaches 90.06% and the value of AUC(area under the receiver curve)reaches 0.945,which can accurately predict the illegal behaviors on the Ethereum trading platform and improve the trading environment based on blockchain.2)In terms of the optimizability of the model constructed by promotion,the gradient lifting book optimization algorithm in the decision tree is used to integrate multiple types of features on the basis of the original attribute features,and the T-SNE algorithm is used to realize the visualization of dimensionality reduction in the process.Multiple cross validation is repeatedly adopted,SHAP Value factor is imported,and different attributes are marked and analyzed.By comparing various other models,the predicted result of the model is 94.29% and the AUC value is 0.9846.The proposed solution can further strengthen the illegal prediction of Ethereum trading platform and continuously improve the security environment of blockchain.3)In order to solve the problem of illegal behavior in Ethereum,this paper puts forward the analysis from the block chain trading network.A central contract account in Ethereum is extracted,the process adopts random sampling method for K-order divergence,and on this basis,a complex feature network is established.The transaction network is constructed by combining with the data set of phish-node accounts extracted from Ethereum.The relevant link index is analyzed by using the network structure of the two,and then it is concluded that the transaction network of Ethereum presents the trend of small-world network.Then,the experimental environment built by Open NE-Pyotch was used to conduct link prediction analysis of the graph representation method.The accuracy of various models was compared.The results of link prediction showed that the graph representation analysis method based on the random walk algorithm had a better score in F1-socre for the fraud prediction existing in Ethereum.In this paper,by extracting a variety of transaction attributes on the blockchain,using optimization algorithms such as machine learning to build a stable model for gradually improving the prediction of illegal transaction behavior of the blockchain,and using complex network theory to build complex characteristic network of the blockchain transaction network,to further explore the analysis of Ethernet smart contract fraud.The experimental results show that the prediction model construction and complex characteristic network construction on the basis of the extracted Ethereum contract account data are gradually and stably improved,which has a better effect on the fraud tracking and analysis of Ethernet smart contract,and the constructed model has better applicability and stability for the future prediction of illegal transactions on Ethernet.