Research On Bitcoin Ponzi Scheme Detection Based On Data Mining Technology

The concept of Bitcoin was proposed by satoshi nakamoto in a paper called Bitcoin: A Peer-to-Peer Electronic Cash System,in which he describes a new type of monetary system,Bitcoin.The bitcoin system is a decentralized monetary system,which does not depend on any third-party institutions(such as the government,financial institutions and other organizations and groups).Everyone can participate in the mining and maintenance of the bitcoin system.Because of this novel idea,bitcoin soon attracted wide attention.In the bitcoin system,the user is anonymous,and the account id of the user is a series of hash values obtained through multiple encryption and hash operations,called address.In the publicly displayed transaction information in the bitcoin network,there is only information such as the amount and time of the transaction between the address and the address,but no specific identity information of the address user,thus ensuring the privacy of the user.However,while anonymity protects the privacy of users,it also provides convenience for some criminals.Since it cannot be connected with the specific identity information of users in the real world,criminals can use the bitcoin system to conduct fraud,money laundering,trafficking in contraband and other illegal activities.Ponzi scheme is a typical type of fraud in the bitcoin system.This fraudulent investment uses the investment of the new investor as a return for the existing investor.When the scheme fails to find a new investor,the whole scheme collapses.The Ponzi scheme in the bitcoin system is endangering a large number of investors,and it is necessary to use data mining technology to detect the bitcoin addresses associated with the Ponzi scheme.The first part of this paper is the introduction part.By explaining the background and significance of the topic and the research situation of domestic and foreign experts and scholars on data mining and analysis in the bitcoin system,the author tried to grasp the current situation and shortcomings of this research field and found the cutting point for the research.The second part is theoretical overview of the bitcoin system,Ponzi scheme and the data mining technology,which mainly defines the bitcoin system,Ponzi scheme,and the data mining,introduces the data structure of the block and the currency trading structure,describes the related mechanism of Ponzi scheme and the operating process,and summarizes the tasks and processes of data mining.The third part is the analysis of Ponzi schemedetection modeling based on data mining technology.Firstly,the author extracted corresponding transaction data according to the address and built trading characteristics to obtain the data sets of Ponzi scheme.Then select SVM,Adaboost and Random Forest three algorithm models to model and analyze the pre-processed data combined with different sampling strategies.According to the comprehensive performance of the model and the existing deficiencies,the model is selected and improved.Based on the third part which illustrates modeling analysis of Ponzi scheme in different sampling strategies and the problem of imbalanced data,on the fourth part,the author improved the Easy Ensemble algorithm and applied to the detection of fraudulent address in Ponzi scheme to improve the detection ability of model for Ponzi scheme.Experimental results show that the improved Easy Ensemble algorithm shows better performance in the detection of Bitcoin Ponzi scheme.