Research And Design Of Ethereum Platform Entity Recognition System

In recent years,block chain and data analysis are entered into the phase of rapid development,especially the block chain field,pouring into the resources not only promoted the block chain itself as a distributed function of financial system,also increased a lot to the research and analysis of data,and as the Ethereum as the birth and rise of 2.0 blocks chain and the most classic account model blockchain platform and related data analysis has brought more technological innovations and challenges,and the development of the block chain also brought many problems,such as user privacy and illegal financial activities,and the characteristics of block chain distributed anonymous public accounting,brings to the data analysis,data is relatively easy to obtain convenient,at the same time it will bring considerable challenges due to anonymize.Therefore,how to de-anonymize blockchain to a certain extent in order to promote the execution of supervision,and how to refine certain information through the data features embodied in the existing data in order to carry out financial regulation has become a very noteworthy issue.In this thesis,the core direction of blockchain de-anonymization,starting with ethereum,focuses on the anonymous information and transaction behavior of blockchain address(account)in account model,with the goal of type identification of blockchain address(account).Aiming at two important processes of data analysis: data preprocessing and data analysis,this thesis proposes three methods for the deanonymization of Ethereum account model,namely heuristic method,tracing source and machine learning recognition.Heuristic methods refer to qualitative identification of user addresses based on intrinsic characteristics of blockchain data or quantifiable unique behaviors of some blockchain users.Tracing and tracing method is based on the core idea that the source of each electronic currency in blockchain is a miner node,and analyzes the source of funds of users so as to complete the definition of the user type.The machine learning method qualitatively identifies users through their unique behavior characteristics that cannot be quantified on the basis of the former two.Then,this thesis applies and engineering the trained model,analyzes the transaction data of blockchain users,identifies the transaction behavior of accounts,quantifies the role and compliance degree of account entities in the blockchain,and finally designs and implements a blockchain entity identification system.Finally,starting from the theoretical process of software engineering,this thesis elaborates and discusses the overall demand analysis,system outline design and module detailed design of the system.The system can realize the function of entity type recognition,that is,the user input address and demand precision,and the system queries the address type and compliance degree through three methods: heuristic method,traceability and machine learning recognition.The account model-based blockchain entity identification system designed and implemented in this thesis can realize the de-anonymization of blockchain to a certain extent.Individual users can avoid certain risks by checking the compliance degree of the trading address;Financial institutions can implement better user management by managing the source and destination of users' funds;Regulators can quantify the degree of compliance of regulated institutions.Therefore,this system can contribute to the development of existing blockchain to a certain extent.