User-Level Fraud Detection Accuracy Using Behavioral Feedback

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing fraud detection systems lack accuracy in determining the fraud level for individual users, leading to potential false positives or false negatives.

Innovation Solution

A fraud detection system that calculates the accuracy of fraud detection for each user by using a learning model to analyze user behavior, determining a fraud level, and comparing it with actual fraud determination results to adjust and improve the fraud level calculation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a fraud detection system calculates fraud levels based on user behavior patterns, then fraud detection capability is provided, but accuracy of fraud detection for individual users cannot be calculated leading to false positives or false negatives

Engineering Contradiction:

Improveaccuracy of fraud detectionVSAvoidlack of accuracy calculation data

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The system implements feedback by storing actual fraud determination results and comparing them with calculated fraud levels. This feedback loop enables the system to calculate accuracy metrics for each user by analyzing the correspondence between predicted fraud levels and actual fraud outcomes, thereby improving measurement precision through continuous learning from real results.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system introduces an intermediary accuracy calculation unit that acts as a mediator between the fraud level calculation process and the actual fraud determination results. This intermediary component calculates and stores accuracy data, enabling the system to measure and improve detection accuracy without disrupting the core fraud detection functionality.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If fraud level calculation is performed for all users, then comprehensive fraud detection is achieved, but system complexity increases

Engineering Contradiction:

Improvefraud detection reliabilityVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system segments the fraud detection process into distinct functional units: a fraud level calculation unit that computes fraud levels, an accuracy calculation unit that measures detection accuracy, and a storage unit that maintains results. This segmentation allows comprehensive fraud detection while managing system complexity through modular design, where each unit has a specific responsibility.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs self-service by automatically calculating accuracy metrics and storing results without requiring external intervention. The accuracy calculation unit autonomously processes fraud determination results and updates accuracy data for each user, reducing operational complexity while maintaining high detection reliability across all users.

Inventive Principle:

Principle #25Self-service

Data Source

PatentEP3882795B1Fraud detection system, fraud detection method, and program

Publication Date: 2026.01.28 RAKUTEN GROUP INC

Figure 1
Figure 2
Figure 3

AI summary

Fraud level calculating means (101) of a fraud detection system (S) calculates a fraud level of a user who uses a service based on a behavior of the user. Determination result obtaining means (102) obtains a determination result as to whether the behavior is actually fraudulent. Accuracy calculating means (103) calculates accuracy of fraud detection for each user based on the fraud level and the determination result.