Purchase Document Entity Tagging with Heuristic Error Correction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing optical character recognition (OCR) techniques often produce erroneous results in complex documents or images, leading to false positives and false negatives in automated entity tagging of purchase documents, particularly for product descriptions, codes, quantities, and prices.

Innovation Solution

Applying post-processing heuristics based on business rules to correct entity tagging errors, including product line grouping and entity linking, using techniques such as GraphNN and heuristics to improve the accuracy of entity tagging in purchase documents.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If traditional AI/ML techniques are used for entity tagging in purchase documents, then automation is achieved, but accuracy deteriorates due to false positives and false negatives

Engineering Contradiction:

Improveautomated entity taggingVSAvoidtagging accuracy

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent segments the entity tagging process into multiple independent stages: initial entity tagging by AI/ML model, heuristic rule application for correction, and verification stages. Each stage handles specific aspects of the tagging process, allowing the system to maintain automation while improving accuracy through structured multi-step processing.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces heuristic rules as an intermediary layer between the AI/ML entity tagging model and the final output. These heuristic rules act as a mediator that corrects false positives and false negatives generated by the AI/ML model, bridging the gap between automated tagging and accurate results without requiring manual intervention.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If AI/ML techniques are applied to correct entity tagging errors, then accuracy is improved, but computational effort and energy consumption increase

Engineering Contradiction:

Improvetagging accuracyVSAvoidcomputational energy

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent employs lightweight heuristic rules that are computationally inexpensive and quick to execute, replacing the need for additional expensive AI/ML model invocations. These simple rule-based corrections provide significant accuracy improvements at minimal computational cost and energy consumption.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent applies heuristic rules immediately after the initial AI/ML entity tagging stage to correct errors before final output generation. This preliminary correction action prevents the need for multiple iterative AI/ML processing cycles, reducing overall computational effort and energy consumption while maintaining high accuracy.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12620249B2Methods, systems, articles of manufacture and apparatus to improve tagging accuracy

Publication Date: 2026.05.05 NIELSEN CONSUMER LLC

AI summary

Methods, apparatus, systems, and articles of manufacture are disclosed to improve tagging accuracy. An example apparatus includes at least one memory, machine readable instructions, and processor circuitry to execute the machine readable instructions to at least search a first row of a document to identify a first row that includes a first type of entity, search the first row of the document to identify a second type of entity that is missing, search the first row of the document to identify a first integer value, and associate the first row with a product corresponding to the first integer value.