Cryptocurrency Terrorist Financing Transaction Behavior Analysis And Address Prediction

The popularization of information technology and Internet has brought new means of terrorist financing to terrorism,that is,using cryptocurrency for terrorist financing.Terrorist financing refers to the financing of terrorism in any way,which is the primary condition for the development of terrorism.The characteristics of cryptocurrency,such as anonymity and low cost,meet the needs of terrorist financing.In the illegal transactions of cryptocurrency,the share of terrorist financing is increasing year by year.However,there is still a lack of academic research on the analysis of cryptocurrency terrorist financing.At present,terrorist organizations mainly use bitcoin for terrorist financing,with the form of publishing bitcoin addresses on social media platforms to raise funds.Therefore,this research takes bitcoin as the research object,and collects public bitcoin terrorist financing case data for transaction behavior analysis and address prediction.Firstly,this paper uses entity modeling and transaction relationship modeling methods to build transaction network data,and uses descriptive statistics and network analysis methods to analyze the bitcoin terrorist financing behavior from three dimensions,namely recruitment address,capital source and capital destination.The results show that in terms of recruitment address analysis,terrorist organizations raise funds through temporary transit address.In terms of funding sources,most of the donor entities are unknown entities.In terms of the whereabouts of funds,the key nodes are identified by Central Rank,which combing Page Rank and centrality analysis indicators.The known terrorist financing entities are found effectively.Furthermore,the abuse exchanges and service providers such as Binance exchange and Huobi exchange are found.By locating the key nodes,this research summarizes three transaction modes of terrorist organization fund transfer: first,the fund collected from the recruitment address directly transfers to the exchange or service provider;second,it transfers to the exchange or service provider through the transit address;third,the exchange or service provider acts as the transit address to disperse the fund and hide the whereabouts.Secondly,based on the characteristics of bitcoin terrorist financing transactions and previous literature analysis,basic transaction features and transaction associated addresses features are used to predict the terrorist financing address of bitcoin.For associated addresses features,it mainly extracts the associated addresses entity labels of the input and output of transaction.This paper uses SVM,Logistic Regression,MLP Random Forest and XGBoost to compare and analyze the models of different feature combinations.The results show that the addition of transaction related addresses features can help to improve the classification effect of basic features model.The ensemble learning model Random Forest gets best performance that F1 value reaches0.89 and AUC value reaches 0.97.The result shows the importance of the features of transaction associated addresses,which indicates the behavior that terrorist organizations tend to trade with unknown addresses or abuse addresses in the financing process.Finally,the paper puts forward suggestions including strengthening the information supervision of social media terrorist organizations,improving the supervision mechanism of exchanges and service providers,and paying attention to the transaction associated addresses features,so as to effectively combat the terrorist financing of cryptocurrency and reduce the harm of terrorism.