User Segmentation System for Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Portable financial device issuing institutions and transaction service providers face challenges in determining users' propensity to make purchases in foreign countries, leading to potential false fraud alerts and missed opportunities for offering travel benefits, due to the lack of efficient methods to segment users based on predicted activity outside their home region.
Innovation Solution
A method and system that segment users in a first region based on predicted activity in a second region by determining user subsets with varying transaction histories, generating activation metrics using algorithms, and automatically initiating target actions for users with a propensity for foreign transactions, such as offering travel benefits or approving transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If transaction monitoring is implemented to detect foreign transactions, then fraud detection capability is improved, but false fraud alerts increase for legitimate travelers
Solution Approach 1:
The patent segments users into different groups based on their travel propensity scores derived from transaction data analysis. By dividing the user base into segments (e.g., high propensity travelers, low propensity travelers), the system can apply different monitoring thresholds and alert strategies to each segment, thereby reducing false alerts for legitimate travelers while maintaining fraud detection for suspicious transactions.
Solution Approach 2:
The system performs preliminary analysis of transaction data to generate travel propensity scores before actual foreign transactions occur. This preliminary action allows the system to pre-identify users who are likely travelers and adjust their monitoring parameters in advance, so that when these users make legitimate foreign transactions, they are not flagged as fraud.
2Reliability
If users must place foreign travel notice before traveling, then fraud protection is improved, but user convenience deteriorates
Solution Approach 1:
The system automatically analyzes transaction data and generates travel propensity scores without requiring user intervention. Users do not need to manually place travel notices; the system self-identifies travelers based on their spending patterns and automatically adjusts monitoring parameters, thereby maintaining fraud protection while eliminating the convenience burden of manual notice placement.
Solution Approach 2:
The system continuously monitors transaction data and provides feedback by updating travel propensity scores in real-time. This feedback mechanism allows the system to automatically recognize when a user is traveling based on their spending behavior and adjust fraud monitoring accordingly, eliminating the need for users to proactively notify the system of their travel plans.
3Measurement precision
If transaction data is analyzed in real-time for all users, then detection accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the user population based on travel propensity scores and applies different levels of monitoring intensity to each segment. High-propensity travelers receive lenient monitoring with higher transaction thresholds, while low-propensity travelers receive stricter monitoring. This segmentation allows the system to maintain detection accuracy for fraud while reducing the overall computational burden by not applying full real-time analysis to all users equally.
Solution Approach 2:
The system dynamically changes monitoring parameters such as transaction thresholds and alert sensitivity based on user-specific travel propensity scores. Instead of using fixed parameters for all users, the system adjusts parameters individually or in segments, thereby maintaining high detection accuracy while reducing system complexity through parameter optimization rather than universal complex analysis.
Data Source
AI summary
A method of segmenting a plurality of users in a first region based on predicted activity external to the first region. A system and computer program product are also provided.