Token-Based Event Indicator Aggregation for Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting identity fraud are inefficient, often taking weeks or months to recognize, and can increase the risk of fraud due to the need for manual account verification and generate many false positives, making it difficult to sustain and effectively combat criminal activity.
Innovation Solution
A computer-implemented method and system that enables the aggregation of event indicators from transaction entities to an aggregating entity using tokens, allowing for real-time monitoring and identification of suspicious activities without exposing sensitive information, thereby reducing the risk of fraud detection delays and false alarms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual account verification is used for fraud detection, then detection accuracy can be maintained, but detection time increases to weeks or months
Solution Approach 1:
The patent introduces an intermediary system that aggregates event indicators from multiple transaction entities and uses machine learning models to automatically detect fraud patterns. This intermediary layer enables real-time analysis without requiring manual verification of each transaction, thus reducing detection time from weeks/months to real-time while maintaining accuracy through sophisticated algorithms.
Solution Approach 2:
The patent replaces the mechanical manual verification process with an automated electronic system that collects, aggregates, and analyzes event indicators using machine learning algorithms. This substitution eliminates the time-consuming manual aspect while preserving detection accuracy through advanced computational analysis of transaction patterns.
2Productivity
If pattern recognition approaches are used for fraud detection, then automated detection can be achieved, but false positive rate increases significantly
Solution Approach 1:
The patent segments the fraud detection process into multiple independent components: event indicator collection from various sources, aggregation of these indicators, and separate machine learning model analysis for different fraud patterns. This segmentation allows each component to be optimized independently, improving automated detection while reducing false positives through multi-factor validation rather than relying on a single pattern recognition system.
Solution Approach 2:
The patent changes the parameters of detection by using multiple event indicators and weighting them differently based on their significance. Instead of relying on a single pattern recognition threshold that causes false positives, the system uses multiple parameters (different event types, frequencies, and combinations) that can be adjusted to optimize detection accuracy and minimize false alarms.
3Measurement precision
If self-police methods are used for account monitoring, then detection accuracy can be maintained, but operational complexity and time consumption increase
Solution Approach 1:
The patent implements a self-service system where the fraud detection operates automatically without requiring user intervention. The system continuously collects event indicators, aggregates them, and applies machine learning models autonomously to detect fraud patterns. This eliminates the complexity of manual self-police methods while maintaining high detection accuracy through automated intelligent analysis.
4Reliability
If manual verification methods are used, then fraud detection can be performed, but user time investment and risk exposure increase
Solution Approach 1:
The patent introduces an intermediary automated system that handles the time-consuming verification tasks. Users simply need to review alerts generated by the system rather than manually verifying each transaction. This intermediary layer performs the heavy lifting of data collection, aggregation, and analysis, reducing user time investment while maintaining detection effectiveness through sophisticated automated analysis.
Data Source
AI summary
A user inspects at least one indicator of an event. The user enables a token corresponding to an account of an aggregating entity to be received by a transaction entity and identifies at least one type of event of interest to be reported by the transaction entity to the aggregating entity. The user obtains and inspects at least one indicator from the account of the aggregating entity, where each obtained indicator is adapted to be created by the aggregating entity based upon an event message received from the transaction entity. The event message comprises the token, which is adapted to be used by the aggregating entity to identify the account and the event message corresponds to an occurrence of an event of at least one type of event of interest to be reported by the transaction entity to the aggregating entity.