Risk User Network Label Propagation for Fraud Detection Timeliness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for recognizing fraudulent users in mobile payment systems are inefficient due to poor timeliness, limited coverage, and low relevance, as they rely on loss reporting and independent user risk labels, failing to comprehensively detect and prevent fraudulent activities across various service scenes.
Innovation Solution
An object recognition method that uses an electronic device to predict risk labels by analyzing user data through an object recognition model, creating a risk user relationship network, and employing label propagation to assess and propagate risk labels across associated users, enhancing timeliness and coverage in fraud detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If loss reporting and independent user risk labels are used for fraud detection, then the system is simple to operate, but the timeliness and coverage of fraud detection are poor
Solution Approach 1:
The system performs preliminary risk assessment by predicting risk labels for users before fraudulent activities occur. The object recognition model continuously analyzes user data and propagates risk labels through the risk user relationship network, enabling proactive detection and prevention of fraud rather than reactive response after loss reporting.
Solution Approach 2:
The risk assessment system is segmented into multiple independent components: data acquisition module, object recognition model, risk user relationship network construction, and label propagation module. This segmentation allows the system to maintain operational simplicity while achieving comprehensive fraud detection through coordinated operation of specialized modules.
2Reliability
If independent user risk labels are used, then the system is easy to implement, but the coverage and relevance of fraud detection are limited
Solution Approach 1:
The risk user relationship network serves as an intermediary structure that connects users through their relationships. The label propagation mechanism uses this network to transfer and refine risk labels, improving detection accuracy by considering contextual relationships rather than treating users in isolation.
Solution Approach 2:
The system implements feedback through iterative label propagation, where predicted risk labels are continuously refined by propagating information through the risk user relationship network. This feedback loop enhances the reliability of fraud detection by repeatedly adjusting predictions based on network relationships and observed patterns.
3Productivity
If traditional loss reporting methods are used, then the system requires minimal data processing, but the timeliness of fraud prevention is poor
Solution Approach 1:
The system replaces manual loss reporting mechanisms with automated machine learning-based risk prediction. The object recognition model automatically analyzes user data and generates risk labels, eliminating the time delay associated with manual reporting while maintaining operational efficiency through automated processing.
Solution Approach 2:
The system performs self-service fraud detection by automatically acquiring user data, predicting risk labels, and propagating risk information through the risk user relationship network without requiring manual intervention. This automation significantly improves both the efficiency and timeliness of fraud detection.
Data Source
AI summary
Embodiments of this application provide an object recognition method performed by an electronic device. The method includes: obtaining relevant object data of target objects; predicting first labels of the various target objects by an object recognition model on the basis of the relevant object data of each target object; obtaining a reference data set comprising relevant object data and second labels of a plurality of first sample objects with annotation labels, and determining first association relationships between the target objects and the plurality of first sample objects; and obtaining recognition results of the target objects according to the first labels of the target objects, the annotation labels and second labels of the first sample objects, and the corresponding first association relationships.