Research On Key Technology Of Anomaly Detection Based On Ethereum Transaction Behavior

There have been scam activities that appeared on Ethereum,which has caused a lot of economic losses and adverse social effects.Due to the anonymity of blockchain,it has increased the difficulty of cracking down on related illegal activities.In order to detect the scams on the Ethereum network in time,especially the Ponzi scheme smart contracts,this paper studies the key technologies from two aspects:feature extraction and algorithm optimization.In terms of feature extraction,the current method for feature extraction of Ethereum addresses lacks internal transaction information,and the feature classification is not clear,which limits the detection performance of the model.This paper firstly aims at the feature extraction problem of Ethereum address transaction information,adds the internal transaction information features,sorts the transaction features into twelve categories,and uses wrapper feature selection method for the extracted features.And this paper further analyzes the transaction characteristics and opcode characteristics for the feature extraction method of the Ponzi scheme contract detection problem.The feature method in this paper improves the AUC value by 2.57%on average compared with the original feature method.The experimental results show that the feature extraction method of the Ethereum address in this paper can better solve the problem of Ethereum Ponzi scheme contract detection compared with the existing research.In terms of algorithm optimization,the current main problem is that the number of Ponzi scheme contracts in the public dataset is small and the positive and negative samples of the training set are unbalanced,which leads to a low recall rate.For low recall of detection models,this paper proposes a cost-sensitive loss function,which can strengthen the learning of the minority Ponzi scheme contract samples and the more confusing samples that are easily misjudged by the model during model iterative training.And this paper proposed a cost-sensitive detection model for Ethereum Ponzi schemes.Compared with the existing algorithm,the algorithm in this paper improves the recall rate by 2.50%and the AUC value by 2.11%.The experimental results show that the cost-sensitive detection algorithm proposed in this paper can better solve the problem of low recall rate of the Ethereum Ponzi scheme detection.