OCR Text Alignment and Tokenization for Key-Value Extraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing systems face challenges in automatically extracting key-value pairs from diverse digital documents, especially when documents have different formats and layouts, requiring manual effort and multiple configurations, and are not efficient for large collections of documents.

Innovation Solution

A system that uses optical character recognition (OCR) to generate text from documents, aligns characters, and tokenizes them to identify keys and values without prior knowledge of document layouts, allowing for clustering and extraction of key-value pairs across multiple documents, including those with tables, using a combination of OCR, aligner, and tokenizer processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual extraction methods are used for different document types, then extraction accuracy is improved, but processing time and labor requirements increase significantly

Engineering Contradiction:
Improveextraction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the document processing task into distinct stages: OCR text extraction, character alignment, tokenization, and key-value pair identification. This segmentation allows each stage to be optimized independently and enables automated processing across different document types without manual intervention.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs self-service by automatically learning document layouts and extracting key-value pairs without requiring manual configuration or training for each document type. The algorithm adapts to different forms and layouts autonomously, eliminating the need for manual extraction methods.

Inventive Principle:
Principle #25Self-service

2Reliability

If multiple configurations are created for different document formats, then extraction reliability is improved, but system complexity increases

Engineering Contradiction:
Improveextraction reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent creates a universal extraction system that handles multiple document formats and layouts through a single unified algorithm. The system learns from diverse documents and automatically adapts its extraction logic, eliminating the need for separate configurations for each document type while maintaining high reliability.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system dynamically adjusts extraction parameters based on the learned document characteristics. By changing parameters such as key-value delimiters, layout patterns, and extraction thresholds according to the specific document type, the system maintains high reliability without requiring manual configuration for each format.

Inventive Principle:
Principle #35Parameter changes

3Ease of operation

If automated extraction without configuration is implemented, then ease of operation is improved, but extraction accuracy may deteriorate

Engineering Contradiction:
Improveease of operationVSAvoidextraction accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The system performs preliminary learning by analyzing document characteristics before extraction begins. Through pre-processing steps including OCR, alignment, and tokenization, the system prepares and understands the document structure in advance, enabling accurate extraction without requiring manual configuration or user intervention.

Inventive Principle:
Principle #10Preliminary action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables efficient and automated extraction of key-value pairs from various documents, including those in image or PDF formats, without requiring specific configurations, facilitating data consolidation and processing in cloud-based storage systems.

Implementation Method 1

uses optical character recognition (OCR) to generate text from documents

Methodology Applied
Scientific EffectOptical character recognition:

Data Source

PatentEP3864527B1Key value extraction from documents
Publication Date: 2024.08.28 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP3864527B1 patent drawingFigure 1
  • EP3864527B1 patent drawingFigure 2
  • EP3864527B1 patent drawingFigure 3

AI summary

Systems, methods, and computer-executable instructions for extracting key value data. Optical character recognition (OCR) text of a document is received. The y-coordinate of characters are adjusted to a common y-coordinate. The rows of OCR text are tokenized into tokens based on a distance between characters. The tokens are ordered based on the x,y coordinates of the characters. The document is clustered into a cluster based on the ordered tokens and ordered tokens from other documents. Keys for the cluster are determined from the first set of documents. Each key is a token from a first set of documents. A value is assigned to each kay based on the tokens for the document, and values are assigned to each key for the other documents. The values for the document and the values for the other documents are stored in an output document.