Probabilistic Matching of Non-Indexed Payment Records
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Issuer institutions face challenges in accurately determining the correspondence between clearing records and authorization records, particularly when the transaction amounts differ or authorization records are removed, leading to potential fraudulent or unauthorized force-post payment transactions.
Innovation Solution
A computer-implemented method and system that compares key fields of clearing records with authorization records, normalizes data, and uses machine learning models to generate updated clearing records with confidence scores, enabling accurate matching and reducing the risk of chargebacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional exact matching methods are used to match clearing records with authorization records, then the matching process is simple and fast, but the accuracy deteriorates when transaction amounts differ or authorization records are removed
Solution Approach 1:
The system transforms the matching approach from exact parameter matching to probabilistic matching by introducing confidence scores. Instead of requiring exact matches on transaction amounts and record identifiers, the system calculates likelihood values that accommodate variations in transaction amounts (e.g., due to tips or currency changes) and handles missing authorization records through probabilistic inference.
Solution Approach 2:
The system introduces an intermediary matching layer that sits between traditional exact matching and final record pairing. This intermediary calculates confidence scores based on multiple factors including transaction amount similarity, timestamp correlation, and merchant information matching, thereby mediating the connection between clearing records and authorization records even when direct matches are not exact.
2Ease of operation
If issuer institutions process clearing records as force-post transactions when matches cannot be determined, then the processing is simplified, but the risk of fraudulent or unauthorized transactions increases
Solution Approach 1:
The system performs preliminary matching attempts using confidence score calculations before defaulting to force-post processing. By pre-calculating the likelihood of matches using multiple data points and comparison algorithms, the system can identify potential matches that would otherwise be missed, thereby reducing unnecessary force-post transactions and associated fraud risks.
Solution Approach 2:
The system implements a feedback mechanism where confidence scores from matching attempts influence subsequent processing decisions. When confidence scores indicate a likely match, the system proceeds with that matching result; when scores are insufficient, the system can adjust processing parameters or request additional verification, thereby creating a feedback loop that improves reliability without completely sacrificing processing efficiency.
3Reliability
If issuer institutions perform extensive matching analysis to reduce fraudulent transactions, then the reliability improves, but the processing time and network resource consumption increase
Solution Approach 1:
The system applies partial matching analysis by focusing on the most discriminating factors first (such as transaction amount ranges, timestamp windows, and merchant identifiers) rather than performing exhaustive comparisons of all possible record attributes. This allows the system to achieve sufficient reliability for fraud detection while avoiding the time and resource costs of complete analysis in every case.
Data Source
AI summary
Systems, computer-implemented methods, and computer program products for determining correspondence of non-indexed records are described herein. The method may include receiving a clearing record including at least one key field, comparing a value associated with a first key field of the clearing record to a value associated with a first key field of one or more authorization records associated with one or more payment transactions, and determining that the clearing record corresponds to an authorization record from among the one or more authorization records based on comparing the value associated with the first key field of the clearing record to the value associated with the first key field of the one or more authorization records. The method also includes generating an updated clearing record based on determining that the clearing record corresponds to the authorization record, and transmitting the updated clearing record.