DFS Graph Mining for Frequent Subgraph Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing risk control systems rely heavily on manual updates of expert rules, making it difficult to timely adapt to rapidly changing fraud situations, especially in identifying new and unknown fraud patterns.
Innovation Solution
A graph mining method that utilizes depth-first search (DFS) codes to extend K-th-order patterns and instances, determining support for all orders of patterns, and identifying frequent subgraphs based on a predetermined threshold, enabling automated and efficient fraud pattern detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual updates of expert rules are used in risk control systems, then known attacks can be well prevented and controlled, but the system cannot timely adapt to rapidly changing fraud situations and new unknown fraud patterns
Solution Approach 1:
The system performs automated graph mining to discover fraud patterns independently without requiring manual intervention. The automated pattern mining process continuously monitors transaction graphs, identifies frequent subgraphs, and generates risk alerts autonomously, enabling the system to adapt to new fraud patterns without manual rule updates
Solution Approach 2:
The patent replaces manual mechanical rule-setting with automated computational graph mining. Instead of manually creating and updating risk rules, the system uses algorithmic pattern discovery (graph mining, frequent subgraph identification) to automatically detect fraud patterns, substituting human expertise with automated computational mechanisms
2Adaptability or versatility
If automated graph mining is implemented to detect fraud patterns, then timely adaptation to changing fraud situations is enabled, but the complexity of the risk control system increases
Solution Approach 1:
The graph mining process is segmented into distinct manageable stages: graph construction from transaction data, pattern mining using graph mining algorithms, support calculation for pattern frequency determination, and risk alert generation. This segmentation allows the complex automated system to be implemented modularly, reducing overall system complexity while maintaining adaptability
3Measurement precision
If manual rule maintenance is performed continuously, then risk control rules remain accurate for known attacks, but a great deal of manual experience and time is required
Solution Approach 1:
The graph mining operation runs continuously on incoming transaction data without interruption. The system continuously builds transaction graphs, mines patterns, and identifies frequent subgraphs in real-time, eliminating the need for periodic manual rule updates while maintaining continuous accuracy in fraud detection
Solution Approach 2:
The system incorporates feedback mechanisms where mined frequent subgraphs and identified patterns are used to generate risk alerts that feed back into the risk control process. This feedback loop enables the system to automatically learn from detected patterns and adjust its detection criteria without manual intervention, maintaining rule accuracy while eliminating manual update time
Data Source
AI summary
A graph mining method includes: obtaining a depth first search (DFS) code corresponding to a to-be-mined full graph in a predetermined scenario; extending a K-th-order pattern and a K-th-order pattern instance of the to-be-mined full graph based on the DFS code, to obtain a (K+1)-th-order pattern and a (K+1)-th-order pattern instance of the to-be-mined full graph, wherein K is an integer greater than or equal to 0; determining, based on quantities of pattern instances corresponding to all orders of patterns of the to-be-mined full graph, support corresponding to all the orders of patterns; and determining a frequent subgraph of the to-be-mined full graph based on the support corresponding to all the orders of patterns.