Analysis Of Bitcoin Network Transaction Characteristics And Research And Design Of Entity Recognition Algorithm

Blockchain is still in a good period of development,but many problems have been exposed during its development,such as being used in illegal transactions such as money laundering and gambling.As the earliest and most representative blockchain system,Bitcoin has experienced the most controversy.Most of these disputes do not deny the technology itself,but stem from the lack of sufficient knowledge and supervision of the Bitcoin network.Bitcoin deanonymization,as a research,is to understand and analyze the development status and evolution process of the Bitcoin network,and for the deanonymization of specific account entities on the blockchain,the identification work can also assist in supervision and investigation and evidence collection.Therefore,it has high research value.In order to better de-anonymize Bitcoin,this article sorts out and analyzes the Bitcoin network from the perspectives of macro network portrayal and micro address entity recognition.The main contributions are as follows:1.Analyze the characteristics of the Bitcoin transaction network from the perspective of a macro complex networkFrom the various indicators of the complex network as the starting point,this article studies the basic indicators of the Bitcoin network,and completes the analysis and verification of the viewpoints after proposing hypotheses based on field experience.It was found that the correlation between the transaction volume and the amount of account funds in the Bitcoin network remained stable.The retention rate of savings accounts has remained in a low range for a long time after 2012.Most of the early savings account users are actually still in the current ecology.Participants in the transaction.2.Propose a method of entity recognition and classification using associated featuresAt present,the mainstream Bitcoin network entity recognition method uses the structural characteristics of the network graph.This thesis starts from the multitransaction correlation characteristics of the Bitcoin address itself as an entry point,and proposes a method to use these correlation characteristics for Bitcoin entity recognition.In this thesis,we crawled tag data from publicly labeled websites to construct a corresponding data set,and verified from experiments that this method can effectively improve the performance of entity recognition.On this basis,this article also analyzes the importance of features and the relevance of classification.3.Design and implement a feature analysis and entity recognition system for characterizing the Bitcoin networkIn response to the above transaction network feature analysis and the actual needs of entity recognition,this thesis also combines the actual machine performance to design a set of highly automated scheduling,which can be interrupted and resumed in time,and has an abnormal alarm function.Bitcoin network feature analysis and entity recognition The system can seamlessly migrate to other Bitcoin networks to serve network analysis and characterization.From the results,the recognition method proposed in this thesis can improve the performance of entity recognition,and the analysis work further deepens our understanding of the Bitcoin network.The realized analysis and identification system can complete the output of the main indicators without manual intervention after the basic configuration,and can be interrupted and restored in time.This system is also very helpful for the research of Bitcoin de-anonymity and the reduction of debugging configuration in the analysis of such large amounts of data.