Graph-Based Fraud Detection System for Banking Networks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods are inadequate in tracing and preventing the rapid dispersion of funds in banking fraud, as they rely on manual investigation and lack real-time identification of fraudulent transactions across multiple bank accounts, making it difficult to recover stolen money and close associated accounts.

Innovation Solution

A system that uses a computer program and apparatus to rapidly identify fraudulent accounts and transactions by building traceable messages through a network of bank accounts, employing graph theory and a Breadth-First traversal method to quickly detect and isolate fraudulent activity, allowing for timely notification and closure of affected accounts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If manual investigation methods are used to trace fraudulent funds, then investigation can be conducted with simple tools, but the tracing speed is too slow and cannot keep up with the rapid transfer of funds

Engineering Contradiction:
Improvetracing speedVSAvoidsystem complexity
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent replaces manual mechanical investigation methods with an automated computer-based system that uses graph theory algorithms to trace fraudulent funds. The system automatically processes transaction data, builds transaction graphs, and identifies fraudulent accounts, eliminating the need for manual tracing while dramatically increasing speed.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs preliminary actions by continuously monitoring and recording transaction data in real-time, building the transaction graph proactively before fraud occurs. When fraudulent activity is detected, the pre-built graph structure enables immediate tracing without requiring data collection during the investigation, thus accelerating the tracing speed.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If funds are transferred through numerous bank accounts rapidly, then the perpetrator can disperse money and make tracing complicated, but this creates a pattern that can be detected through network analysis

Engineering Contradiction:
Improvefraud detection accuracyVSAvoidtracing difficulty
Core Design Contradiction:
ReliabilityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent transitions from traditional linear, bank-by-bank tracing to a multi-dimensional network analysis approach. By representing transactions as a graph with nodes (accounts) and edges (transactions), the system can analyze relationships across multiple dimensions simultaneously, detecting patterns of fund dispersion that would be invisible in traditional sequential analysis.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The transaction graph system serves multiple functions: it traces individual fraudulent transactions, identifies networks of connected accounts, detects patterns of fund dispersion, and prioritizes investigation targets. This universal approach handles various fraud scenarios uniformly, improving detection reliability across different fraud types.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Loss of time

If real-time identification of fraudulent transactions is implemented, then money can be stopped and accounts closed quickly, but this requires processing large volumes of transaction data at high speed

Engineering Contradiction:
Improvetime to stop fraudulent moneyVSAvoiddata processing capacity
Core Design Contradiction:
Loss of timeVSProductivity

Solution Approach 1:

The patent segments the transaction data processing into manageable graph operations. By dividing the transaction network into connected components and using efficient graph traversal algorithms, the system can process large volumes of data incrementally rather than requiring monolithic processing, enabling real-time fraud detection without overwhelming computational resources.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3629273B1An apparatus, computer program and method
Publication Date: 2022.03.02 IPCO 2012 LTD
  • EP3629273B1 patent drawingFigure 1
  • EP3629273B1 patent drawingFigure 2A
  • EP3629273B1 patent drawingFigure 2B

AI summary

A method of building a set of traceable messages through a network of nodes is provided, the method comprising receiving message information corresponding to a first message, the message information comprising a first source identifier and a first destination identifier and determining whether the source identifier is associated with a set of previous messages, whereby when the first source identifier is associated with a set of previous messages, the method comprises identifying the first message as also belonging to the set of previous messages and when the first source identifier is not associated with a set of previous messages, the method comprises identifying the first message as belonging to a new set of messages, the method further comprising associating the first destination identifier with the set of messages to which the first message belongs, such that if a second message is sent from that destination, the second message will be identified as belonging to that set of messages.