Cryptocurrency Address Risk Diagnosis Using Clustered Graph And Text Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blockchain-based cryptocurrency systems face challenges in accurately detecting illicit transactions due to the reliance on historical labeled data and the unreliability of public address reports, which are prone to mistakes and spam, hindering real-time illicit activity detection.
Innovation Solution
A fully automated system using unsupervised learning and machine learning algorithms, such as K-means clustering and Graph Neural Networks, processes transaction graphs and public reports to assign risk levels to cryptocurrency addresses, enabling real-time risk diagnosis and prevention of high-risk transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If public address reports are used for labeling, then real-time detection capability is improved, but reliability deteriorates due to mistakes and spam
Solution Approach 1:
The patent introduces an intermediary verification mechanism where multiple public reports are aggregated and cross-validated before being accepted as labels. The system uses a confidence scoring mechanism that only accepts labels meeting threshold requirements, filtering out spam and unreliable reports while maintaining real-time processing capability.
Solution Approach 2:
The system implements feedback loops where detection results are continuously verified against new public reports and transaction patterns. Misclassified transactions trigger re-evaluation and model retraining, creating a self-correcting system that improves reliability over time while maintaining real-time operational capability.
2Measurement precision
If historical labeled data is used for training, then model accuracy is improved, but real-time adaptability deteriorates
Solution Approach 1:
The patent implements a dynamic system where the detection model is continuously retrained using newly verified public reports and emerging transaction patterns. The system adapts its parameters and thresholds in real-time based on current blockchain activity, allowing it to maintain high accuracy while responding to new illicit techniques as they emerge.
Solution Approach 2:
The system performs preliminary clustering and risk level assignment on incoming transactions using pre-trained models, then rapidly refines these assignments using real-time public reports. This two-stage approach allows the system to maintain real-time processing speed while incorporating continuous learning from historical and current data.
3Reliability
If manual verification of public reports is performed, then label reliability is improved, but automation level deteriorates
Solution Approach 1:
The system implements self-service automation where the detection model automatically verifies public reports against transaction graphs and existing labels, assigning confidence scores without human intervention. The system self-corrects by identifying and discarding contradictory or spam reports, maintaining high reliability while operating fully automatically.
Solution Approach 2:
Automated feedback mechanisms continuously evaluate the quality of public reports by cross-referencing with transaction data and detection outcomes. The system automatically adjusts its acceptance criteria and retraining parameters based on this feedback, maintaining reliability without requiring manual verification while preserving full automation.
Data Source
AI summary
Systems and methods for risk diagnosis of cryptocurrency addresses on blockchains using anonymous and public information are disclosed. In one embodiment, a method may include a risk diagnosis computer program executed by a server: receiving data with labels and data without labels from public data databases and/or anonymous data databases; fitting the data without labels into a plurality of clusters using unsupervised learning; assigning a risk level to each of the plurality of clusters; verifying the risk level for each of the plurality of clusters using the data with labels; receiving unseen data; and predicting a risk level associated with the unseen data by constructing features and text embeddings from the unseen data, clustering the unseen data based on a distance measurement to one of the plurality of clusters, and returning a risk level associated with the cluster.