Frequent-Behavior Sorted Lists for Real-Time Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods fail to effectively detect fraudulent transactions in a timely manner, particularly in telecommunications and financial systems, as fraudsters adapt their methods to evade detection, leading to significant losses for institutions.
Innovation Solution
The implementation of a frequent-behavior sorted list method that dynamically tracks and analyzes patterns of usage for phone numbers, ATMs, and Point of Sale merchants, identifying deviations from normal behavior to flag potential fraud, utilizing a fraudster signature database for comparison and real-time prediction of fraudulent activities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional fraud detection methods are used, then fraud detection capability is maintained, but computation burden increases and real-time detection is not achieved
Solution Approach 1:
The patent segments the fraud detection process by maintaining separate frequent-behavior sorted lists for different entities (called numbers, ATMs, merchants) and using independent counters for each entity. This segmentation allows parallel processing and reduces the computational complexity of analyzing all transactions comprehensively, enabling real-time fraud detection with reduced computation burden.
Solution Approach 2:
The patent performs preliminary action by pre-computing and maintaining frequent-behavior sorted lists that identify commonly accessed entities before fraud analysis is needed. These pre-established lists serve as reference data structures that enable rapid comparison against current transactions, eliminating the need for complex real-time computations and achieving real-time detection with minimal computational overhead.
2Measurement precision
If fraud detection accuracy is improved to detect adapted fraud methods, then detection precision increases, but false positive rate increases and customer inconvenience worsens
Solution Approach 1:
The patent applies local quality by examining specific local patterns in transaction behavior rather than applying uniform detection rules. It analyzes the frequency and recency of access to specific entities (called numbers, ATMs, merchants) and compares these local behaviors against established frequent-behavior patterns. This localized analysis improves detection accuracy for actual fraud while reducing false positives by considering the specific context of each transaction rather than applying blanket restrictions.
Solution Approach 2:
The patent implements dynamics by maintaining dynamic frequent-behavior sorted lists that are continuously updated as new transactions occur. The system adapts to changing user behaviors and fraud patterns by adjusting the frequency thresholds and recency windows. This dynamic approach allows the system to distinguish between legitimate behavior changes and actual fraud, improving detection accuracy while reducing false positives caused by rigid detection rules.
3Reliability
If real-time fraud detection is implemented, then fraud loss is reduced, but system response time requirements increase processing load
Solution Approach 1:
The patent uses copying by creating simplified reference copies of transaction patterns in the form of frequent-behavior sorted lists. Instead of analyzing the full complexity of all historical transactions in real-time, the system maintains compact copied representations of frequent entities with their access frequencies and recency information. These copied structures enable rapid real-time comparison against current transactions, reducing processing load while maintaining the ability to detect fraud patterns and reduce losses.
Data Source
AI summary
A computerized method for detecting fraud includes obtaining frequency information on entities in transaction data for at least one individual account, converting frequency information to a frequency variable, and predicting whether an activity is fraudulent in response to the frequency variable. In some embodiments, the frequency variable is used with at least one other variable to predict fraudulent activity.