Tokenized Account Attributes for Lower-False-Positive Fraud Checks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fraudulent transactions are increasing, and existing fraud detection solutions using predictive AI technology lead to high false positives, impacting resources and blocking legitimate transactions.
Innovation Solution
A system that generates tokens for accounts associated with authorizing entities, assigns transaction rules and attributes, and checks these attributes against rules to determine transaction validity, transmitting validation messages based on rule satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predictive AI technology is applied to flag risky transactions, then fraud detection capability is improved, but false positives increase blocking legitimate transactions
Solution Approach 1:
The system segments account attributes into discrete tokenized attributes (e.g., account age, transaction velocity, geographic location) that can be independently evaluated against transaction rules. This segmentation allows for more precise fraud detection by evaluating specific attribute combinations rather than relying on monolithic AI predictions, thereby reducing false positives while maintaining fraud detection capability.
Solution Approach 2:
The system changes the detection parameters from AI probability scores to deterministic attribute-value comparisons against transaction rules. By transforming continuous AI predictions into discrete attribute evaluations (e.g., 'account age > 30 days' instead of 'fraud probability > 0.7'), the system achieves more reliable and explainable fraud detection with fewer false positives.
2Reliability
If multiple fraud detection solutions are applied, then fraud detection capability is improved, but resource consumption increases
Solution Approach 1:
The system merges multiple fraud detection approaches into a unified attribute-based evaluation framework. Instead of running separate AI models and rule engines independently, the system combines attribute tokenization, transaction rule evaluation, and risk scoring into a single integrated process, reducing redundant processing and lowering overall resource consumption while maintaining comprehensive fraud detection capability.
Solution Approach 2:
The system performs preliminary attribute tokenization and risk scoring before final transaction authorization. By pre-evaluating account attributes against transaction rules in advance, the system identifies high-risk transactions early in the process, allowing for more efficient resource allocation and reducing the processing power required for subsequent fraud detection operations.
3Reliability
If AI technology is used to detect fraud, then fraud detection capability is improved, but false positives consume resources to remedy
Solution Approach 1:
The system incorporates feedback mechanisms where transaction outcomes and attribute evaluations are continuously monitored and used to refine future risk assessments. By learning from actual transaction results and adjusting attribute thresholds accordingly, the system reduces false positives over time, thereby decreasing the resources needed for false positive remediation while maintaining high fraud detection capability.
Data Source
AI summary
Techniques described herein include generating a first token for a first account associated with a first authorizing entity computer, the first token associated with a first attribute. The techniques further include generating a second token for a second account associated with a second authorizing entity computer. The techniques further include assigning a first transaction rule to the second token. The techniques further include assigning a first attribute value to the first attribute associated with the first token. The techniques further include receiving a request to check the first attribute value of the first attribute against the first transaction rule. The techniques further include responsive to the first attribute value of the first attribute satisfying the first transaction rule, transmitting a first validation message to the second authorizing entity computer.