Document Processing Platform Using Machine Learning Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current document processing techniques are expensive, time-consuming, and inefficient, leading to poor customer service, inaccuracies, and wastage of computing resources due to the manual and error-prone conversion of paper-based documents into digital formats.

Innovation Solution

A processing platform utilizing machine learning models, position-based extraction, and automated data labeling to convert image-based documents into digital formats by identifying fields, generating label data, and verifying character strings with confidence scores.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual document processing techniques are used to convert paper-based documents into digital formats, then processing can be performed with simple equipment, but processing time is excessive and accuracy is poor

Engineering Contradiction:
Improveprocessing speedVSAvoidprocessing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical document processing with an automated system combining OCR technology, machine learning models, and computer vision algorithms. The system automatically performs document scanning, text recognition, field extraction, and data validation, eliminating the need for manual handling and significantly reducing processing time while improving accuracy.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The document processing system performs self-service by automatically capturing images, recognizing text, extracting relevant fields, validating data against schemas, and generating digital outputs without human intervention. The system includes self-correction capabilities through confidence scoring and automated validation, allowing it to process documents independently and efficiently.

Inventive Principle:
Principle #25Self-service

2Reliability

If manual document processing is performed, then equipment complexity is low, but processing accuracy and reliability are poor

Engineering Contradiction:
Improveprocessing accuracyVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the document processing task into distinct modular components: image capture, OCR text recognition, field extraction using machine learning models, data validation against schemas, and digital output generation. Each module handles a specific aspect of processing, improving overall reliability through specialized processing while maintaining manageable system complexity through clear separation of functions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system introduces intermediary components including confidence scoring mechanisms, data validation schemas, and error correction modules that mediate between raw OCR output and final digital documents. These intermediaries verify and validate processed data, ensuring high accuracy and reliability while providing a structured approach to managing system complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If automated processing systems are implemented, then processing efficiency improves, but computing resource consumption increases

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidcomputing resource usage
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent implements partial processing by applying machine learning models and validation only to extracted fields that require verification, rather than processing entire documents uniformly. The system uses confidence scoring to determine which fields need manual review versus automatic acceptance, optimizing computing resource usage by focusing intensive processing only where necessary while maintaining high processing efficiency for routine documents.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3882814B1Utilizing machine learning models, position-based extraction and automated data labeling to process image-based documents
Publication Date: 2025.06.25 ACCENTURE GLOBAL SOLUTIONS LTD
  • EP3882814B1 patent drawingFigure 1A
  • EP3882814B1 patent drawingFigure 1B
  • EP3882814B1 patent drawingFigure 1C

AI summary

A device may receive image data that includes an image of a document and lexicon data identifying a lexicon, and may perform an extraction technique on the image data to identify at least one field in the document. The device may utilize form segmentation to automatically generate label data identifying labels for the image data, and may process the image data, the label data, and data identifying the at least one field, with a first model, to identify visual features. The device may process the image data and the visual features, with a second model, to identify sequences of characters, and may process the image data and the sequences of characters, with a third model, to identify strings of characters. The device may compare the lexicon data and the strings of characters to generate verified strings of characters that may be utilized to generate a digitized document.