OCR Line Mapping for Structuring Unstructured Text Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high cost and inefficiency in managing unstructured documents due to diverse document formats and terms require improved methods for effective operation and management, particularly in industrial settings.
Innovation Solution
A text data structuring method and apparatus using Optical Character Recognition (OCR) to extract and classify text, apply natural language processing for tagging and mapping, and correct misrecognition through similarity calculations and training data analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple repetitive input process is used for unstructured documents, then ease of operation is improved, but productivity deteriorates due to high cost and inefficiency
Solution Approach 1:
The patent replaces manual mechanical input processes with an automated image processing system that uses optical character recognition (OCR), line information extraction, and natural language processing models to automatically extract, classify, and structure text data from unstructured documents, thereby eliminating repetitive manual labor while maintaining operational simplicity
Solution Approach 2:
The system enables self-service processing by automatically extracting text, identifying line information, classifying content through trained models, and structuring data without requiring manual intervention, allowing the system to process unstructured documents independently and efficiently
2Measurement precision
If manual text extraction and classification is performed, then measurement precision is improved, but loss of time increases due to repetitive processing
Solution Approach 1:
The patent substitutes manual text extraction and classification with automated optical character recognition (OCR) technology and trained natural language processing models that can accurately identify text content, line information, and document structure, achieving both high precision and time efficiency simultaneously
3Adaptability or versatility
If diverse document formats are handled manually, then adaptability is improved, but device complexity increases due to multiple processing requirements
Solution Approach 1:
The patent implements a universal image processing system that can handle multiple document formats and types through a single integrated framework, where the same OCR and natural language processing models automatically adapt to different document structures, eliminating the need for separate processing systems for each format
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
Enhances the efficiency of operation management by accurately structuring and correcting misrecognized text in unstructured documents, reducing operational costs and improving data quality.
Implementation Method 1
extract text included in an image and position information of the text on the basis of Optical Character Recognition (OCR)
Implementation Method 2
classify the text into text units by applying the text to a natural language processing model that is trained in advance
Implementation Method 3
calculate a first similarity score representing meaning similarity between the first text and the third text and a second similarity score representing meaning similarity between the second text and the third text
Data Source
AI summary
A text data structuring apparatus includes a processor, wherein the processor is configured to: extract text included in an image and position information of the text on the basis of Optical Character Recognition (OCR), extract line information included in the image by using the text, the position information, and the image, classify the text into text units by using natural language processing model, add a tag that combines a first tag and a second tag to each of the keywords included in each of the text units, connect keywords with related tags among the tagged keywords, label each of the connected keywords with key or value, acquire a mapping candidate group based on the line information on the basis of the line information, and decide which texts to be mapped to each other based on similarity scores between texts included in the mapping candidate group.


