Research On Ponzi Scheme Smart Contract Detection Method For Digital Currency

In recent years,the rapid development of blockchain technology has promoted the largescale application of digital currency,and decentralized finance(De Fi)has also developed rapidly.In order to ensure the health of the decentralized financial ecosystem,it is essential to reduce the transaction risk in the system.As a representative De Fi ecosystem platform,Ethereum faces serious fraud threats in the transaction process due to its anonymity.The smart Ponzi scheme is a typical fraud,which causes millions of dollars of losses to investors every year and seriously affects the healthy development of Ethereum.In the context of the current digital economy,the research on fraud detection on Ethereum has just risen,and the detection of the smart Ponzi scheme is the most important.Timeliness is an inevitable requirement for Ponzi scheme detection.It is necessary to detect and disclose Ponzi scheme before consumers are cheated.At the initial stage of contract creation,Ponzi fraud was not obvious and was difficult to be detected.As more and more contracts are called and more and more transactions are involved in the contract,the Ponzi fraud will gradually show signs.Compared with the total number of smart contracts on Ethereum,the number of Ponzi fraud contracts marked accounts for a small proportion.The distribution of Ponzi scheme contracts and Non-Ponzi scheme contracts shows the characteristics of unbalanced categories,which puts forward high requirements on the performance of the detection model.At present,most of the existing intelligent Ponzi fraud detection methods use machine learning algorithms to build detection models based on abnormal Ponzi fraud smart contract samples.However,in general,previous studies did not preprocess the category imbalance problem in the sample dataset.The selection of detection features is not comprehensive enough,and the detection accuracy needs to be further improved.In this paper,based on the verified smart contract address,the smart contract code features and sequential transaction features are extracted to build a detection dataset.The adaptive comprehensive oversampling technology is used to expand the negative Ponzi fraud samples in the data set to ensure the classification of the data set.The long short-term memory neural network is used to build a smart contract Ponzi fraud detection method TTPS that considers time series transaction information,and on the basis of the experiment,the supervision suggestions for the smart Ponzi fraud are given.The main work of this paper is as follows:First,this paper analyzes the fraud mechanism of Ponzi scheme smart contract,and capture the contract bytecode information and transaction information based on the verified address information of smart contract(whether it has been marked as Ponzi scheme or not).Through the corresponding relationship between the byte code and the operation code on the Ethernet yellow page,the contract byte code is converted into human readable operation code and the account features are extracted.The transaction information involved in the contract is sorted,the temporal account features are extracted,and the detection feature system of this paper is constructed by combining the account features and code features.Second,Adaptive Synthetic Sampling(ADASYN)technology is used to deal with the problem of data category imbalance in the detection of Ponzi hoax on Ethereum.Around the original Ponzi scheme samples,new Ponzi scheme samples are generated according to different sampling weights.The number of two types of smart contracts in the final dataset is balanced,ensuring the reliability of the experimental results.Third,based on the constructed detection feature system,this paper proposes a smart Ponzi scheme deception model TTPS based on long short-term memory neural network.The advantages of TTPS method are verified by comparison with other five typical detection methods.Based on the experience and conclusion in the construction of smart Ponzi scheme detection model,the application countermeasures and suggestions for Ethereum and digital currency regulatory authorities to prevent Ponzi scheme are given.