A blockchain address classification method and device

By generating transaction vectors of blockchain addresses and using word vector technology and extreme gradient boosting algorithm to train classifiers, the accuracy and resource cost issues of blockchain address classification are solved, and efficient and reliable blockchain address classification is achieved.

CN114911925BActive Publication Date: 2025-10-03THE PEOPLES BANK OF CHINA DIGITAL CURRENCY INST
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202110185440.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-02-10
Publication Date
2025-10-03
Estimated Expiration
2041-02-10

AI Technical Summary

Technical Problem

Existing technologies have difficulty in accurately classifying blockchain addresses. In particular, traditional statistical methods and machine learning methods suffer from problems such as poor classification accuracy, unreliable results, insufficient algorithm generalization, and high hardware resource and time costs.

Method used

By extracting the transaction records of the blockchain address to generate transaction vectors, using word vector generation algorithms such as word2vec to generate sparse matrices, calculating the address vectors, and using the extreme gradient boosting algorithm to train the classifier, the blockchain address can be classified.

Benefits of technology

It improves the accuracy of blockchain address classification, reduces dependence on the external environment, overcomes the error transmission problem, and reduces hardware resources and time costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114911925B_ABST
    Figure CN114911925B_ABST
Patent Text Reader

Abstract

The present invention discloses a blockchain address classification method and apparatus, relating to the field of computer technology. A specific implementation of the method includes generating a transaction vector for each transaction record corresponding to a blockchain address in a blockchain address set, generating a corresponding address vector based on the transaction vector corresponding to each blockchain address to be classified, inputting the blockchain address vector to be classified into a classifier, and determining the category of the blockchain address to be classified. This implementation enables the generation of equal-length vectors representing blockchain addresses for classification by the classifier, effectively expressing the transaction behaviors and relationships of each participating entity in the blockchain. It also avoids the vulnerability of feature engineering in traditional machine learning to external environmental influences, resulting in high classification accuracy and reliable results. It also overcomes the drawbacks of existing methods, such as error propagation issues, insufficient algorithm generalization, and high hardware resource and time requirements.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • User clustering method based on bitcoin transaction time sequence similarity

    CN111104571A

  • Bitcoin address classification method based on improved random forest

    CN111754345A