Multi-Dimensional Fraud Detection via User Abandonment Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current transaction fraud systems monitor card activity at an individual card level, leading to missed fraudulent activities and inadvertently declined valid transactions, resulting in poor consumer experience and lost revenue for merchants and financial institutions.

Innovation Solution

A multi-dimensional fraud detection system that analyzes historical user data across multiple payment vehicles and contextual factors to determine a user-specific abandonment score and institutional risk tolerance, balancing fraud risk against transaction completion likelihood.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If transaction fraud systems monitor card activity at an individual card level, then fraud detection capability is improved, but false declines increase and consumer experience deteriorates

Engineering Contradiction:

Improvefraud detection capabilityVSAvoidtransaction authorization accuracy

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent combines multiple payment vehicle data into a unified user-level profile, merging transaction histories from different cards into a single analytical framework. This allows the system to evaluate fraud risk based on aggregate user behavior patterns rather than isolated card-level data, reducing false declines while maintaining detection capability.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system transitions from one-dimensional card-level monitoring to multi-dimensional user-level analysis by incorporating additional dimensions such as transaction context, user behavior patterns, and cross-payment-vehicle relationships. This dimensional expansion enables more nuanced fraud detection that distinguishes between legitimate and fraudulent activities more accurately.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Reliability

If suspicious transactions are declined without confirmation of fraud, then fraud risk is reduced, but consumer experience and revenue opportunities are lost

Engineering Contradiction:

Improvefraud risk reductionVSAvoidtransaction completion rate

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system dynamically adjusts authorization decisions based on computed user-specific parameters such as fraud risk scores and abandonment scores. Rather than applying fixed decline rules, the system modifies authorization parameters in real-time based on user behavior analysis, allowing legitimate transactions to proceed while blocking fraudulent ones.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system implements feedback loops where transaction outcomes and user responses are continuously monitored and fed back into the scoring models. This enables the system to learn from actual user behavior patterns, refining its ability to distinguish between users who will abandon transactions versus those who will switch payment methods, thereby reducing unnecessary declines.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If multi-dimensional user data is analyzed for fraud detection, then transaction authorization accuracy is improved, but system complexity increases

Engineering Contradiction:

Improvetransaction authorization accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The complex analysis system is segmented into distinct functional modules: data collection components, scoring computation components (fraud risk score, abandonment score), and authorization decision components. Each module handles specific aspects of the analysis independently, making the overall complex system manageable and maintainable while preserving its analytical capabilities.

Inventive Principle:

Principle #1Segmentation

4Measurement precision

If user transaction history across multiple payment vehicles is monitored, then fraud detection accuracy is improved, but user privacy concerns increase

Engineering Contradiction:

Improvefraud detection accuracyVSAvoiduser privacy

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The system introduces an intermediary analytical layer that processes user data through computed scores (fraud risk score, abandonment score) rather than directly exposing or storing detailed personal information. This intermediary layer enables accurate fraud detection while maintaining user privacy by working with aggregated behavioral metrics rather than raw personal data.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20230075970A1Systems and methods for a context-driven electronic transactions fraud detection

Publication Date: 2023.03.09 WORLDPAY LLC

AI summary

Systems and methods are disclosed for establishing a multi-dimensional fraud detection system and payment analysis. One method includes: receiving transaction history of a user, the transaction history including a first payment vehicle and a second payment vehicle; determining, of the received transaction history, one or more instances of switching from one the first payment vehicle to the second payment vehicle; and determining a user-specific abandonment score for the user, based on the determined instances of switching from the first payment vehicle to the second payment vehicle.