Parallel OCR Engine Merging for Insurance Document Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems are inefficient in processing large volumes of complex business information, particularly in the insurance industry, due to the need for manual review and the limitations of existing Optical Character Recognition (OCR) technologies, which struggle with documents of varying quality.
Innovation Solution
A system that combines multiple OCR engines with machine learning models to enhance data processing efficiency, using orchestration and human-in-the-loop tasks to classify and annotate documents, and predictive models to estimate insurance risks based on open and closed claims data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple OCR engines are used to process documents, then accuracy of character recognition is improved, but device complexity and processing time increase
Solution Approach 1:
The patent segments the document processing task by dividing it into multiple parallel OCR engine executions, each handling the same document independently. The results from multiple engines are then combined through result merging logic that selects the most accurate transcription, effectively segmenting the complexity into manageable parallel tasks rather than a single complex process
Solution Approach 2:
The patent merges the output results from multiple OCR engines by comparing their transcriptions and selecting the most accurate one. This merging process combines the strengths of different engines while filtering out errors, achieving higher overall accuracy without requiring each individual engine to be perfect
2Productivity
If multiple OCR engines are deployed in parallel, then processing speed is improved, but resource consumption and system complexity increase
Solution Approach 1:
The patent implements partial parallelism by processing only the text content through multiple OCR engines while handling other document elements sequentially. Additionally, the system uses excessive action by deploying more OCR engines than strictly necessary, relying on the merging logic to filter results, which allows for faster processing while maintaining resource efficiency through intelligent result selection
3Measurement precision
If human-in-the-loop tasks are implemented for quality assurance, then data accuracy is improved, but processing time and operational complexity increase
Solution Approach 1:
The patent performs preliminary automated processing through multiple OCR engines and result merging before human review. This preliminary action pre-processes the document and identifies potential errors, so that human reviewers only need to verify and correct specific problematic areas rather than reviewing entire documents, significantly reducing the time loss from human involvement
4Measurement precision
If comprehensive document classification and routing workflows are implemented, then data processing accuracy is improved, but system complexity and implementation difficulty increase
Solution Approach 1:
The patent applies local quality by implementing different processing workflows and OCR engine selections based on specific document types and characteristics. Rather than using a uniform complex system for all documents, the system tailors the processing approach to each document's specific needs, applying complexity only where necessary to achieve accurate results for that particular document type
Data Source
AI summary
Techniques including a system and method for optical character recognition. The techniques may involve the use of a system. The system may include a plurality of optical character recognition engines configured to process, in parallel, at least one document or portion thereof, and produce output results for each of the optical character recognition engines. The system may include a component adapted to combine the output results of each of the optical character recognition engines and produce a single unified view of the at least one document or portion thereof.