Machine-Learning Wallet Graphs for Digital Asset Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to efficiently and accurately track digital assets used in crimes due to their complex transfer paths and the increase in human error, especially when assets are divided and transferred through repetitive behaviors like wash trading, leading to delayed detection and reduced accuracy.
Innovation Solution
A method utilizing a machine learning model to infer transactions and output nodes and edges on a user interface, allowing users to track digital asset transfers by inputting features and determining exchange hot wallets, with features like transaction frequency and time contributing to the inference process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual tracking methods are used to monitor digital asset transfers, then users can track transactions, but the complexity of transfer paths and repetitive behaviors like wash trading lead to increased human error and reduced tracking accuracy
Solution Approach 1:
The patent replaces manual tracking mechanisms with an automated machine learning system that uses trained models to analyze transaction patterns, identify wash trading behaviors, and track digital asset flows automatically. This substitution eliminates human error while handling the complexity of multi-step transfer paths and repetitive transaction patterns.
2Productivity
If traditional tracking methods are applied to complex digital asset transfers, then all transactions can be monitored, but the time required for tracking increases significantly and early crime detection is delayed
Solution Approach 1:
The patent implements preliminary action by pre-training machine learning models with extensive transaction data before actual tracking begins. The models are prepared in advance to recognize crime patterns, wash trading behaviors, and suspicious transaction flows, enabling immediate and accurate detection without time-consuming analysis when crimes occur.
Solution Approach 2:
The system creates simplified representations (copies) of complex transaction paths through graph visualizations and aggregated statistics. Instead of manually analyzing every individual transaction in complex multi-step transfers, the system generates condensed visual copies showing key transfer routes, wallet connections, and suspicious patterns, dramatically reducing analysis time while maintaining detection accuracy.
3Ease of operation
If users with insufficient blockchain knowledge attempt to track digital assets, then they can access tracking functionality, but they cannot effectively understand or interpret the complex transfer paths
Solution Approach 1:
The patent introduces machine learning models and visualization interfaces as intermediaries between users and complex blockchain data. The system automatically processes raw transaction data, identifies meaningful patterns, and presents simplified visual representations showing transfer paths, wallet relationships, and suspicious activities. This intermediary layer shields users from underlying blockchain complexity while delivering actionable tracking insights.
Data Source
AI summary
A method for providing a tracking result of a digital asset, performed by at least one processor, may include acquiring information associated with a first transaction that corresponds to a transfer of a first target asset from a first wallet to a second wallet, outputting a first node corresponding to the first wallet, a second node corresponding to the second wallet, and a first edge corresponding to the first transaction, outputting a second edge corresponding to a second transaction among one or more transactions that correspond to one or more transfers of an asset from the second wallet to a different wallet, and outputting a third node corresponding to a third wallet that is a recipient of the second transaction, wherein the second transaction is determined to correspond to a transfer of a second target asset associated with at least a portion of the first target asset.