Machine-Learned MCC Classification from Cleansed Transactions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional financial systems fail to accurately classify merchant category codes (MCCs) due to incorrect or outdated codes provided by merchants, leading to inaccurate transaction summaries and reward systems, as well as incorrect fee charging.

Innovation Solution

An entity classification system that uses a machine classifier to determine a recommended MCC based on merchant names and locations, updating the codes to a standardized format like ISO 18245, and generating transaction summaries with corrected data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If merchants provide entity codes manually, then the process is simple and quick, but the accuracy and standardization of the codes deteriorate

Engineering Contradiction:

Improvespeed of code assignmentVSAvoidaccuracy of entity code

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system automatically performs entity code classification using machine learning models and transaction data, eliminating the need for manual merchant input. The algorithm independently determines the appropriate entity code based on analyzed features, making the system self-sufficient in the classification task.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical classification processes with automated computational systems. Machine learning algorithms and natural language processing substitute human judgment, transforming the entity code assignment from a manual administrative task to an automated data processing operation.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Speed

If conventional systems use provided entity codes directly, then processing is fast, but the reliability of transaction data deteriorates

Engineering Contradiction:

Improveprocessing speedVSAvoidaccuracy of transaction data

Core Design Contradiction:

SpeedVSReliability

Solution Approach 1:

The system performs preliminary entity code verification and correction before final transaction processing. By pre-analyzing transaction data and validating entity codes against established criteria, the system ensures data reliability is established upfront, preventing errors from propagating through subsequent processing stages.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback mechanisms where transaction data is continuously analyzed to improve entity code accuracy. The system learns from patterns in transaction data and adjusts classifications accordingly, creating a closed-loop system that enhances reliability over time while maintaining processing efficiency.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20250284673A1Entity Classification Using Cleansed Transactions

Publication Date: 2025.09.11 CAPITAL ONE SERVICES LLC

AI summary

Systems as described herein may classify entities based on cleansed transactions. An entity classification server may obtain transaction data indicating an entity name and an entity code in a non-standardized format. A recommended entity code in a standardized format may be determined from a remote data store. The entity classification server may generate a score indicating a likelihood that the recommended entity code correctly identifies the entity indicated in the transaction data using a machine classifier. The entity classification server may update the entity code in the transaction data with the recommended entity code based on the score exceeding a threshold value. Accordingly, a transaction summary comprising the transaction data may be generated and provided to a computing device.