Key-Value Extraction from Documents via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for extracting data from documents are inefficient and resource-intensive, requiring significant manual effort and being prone to errors, especially when dealing with documents of varying layouts and content.
Innovation Solution
A machine learning-based technique that involves determining a set of region proposals bounding text in documents, applying a machine learning model to generate predictions for key-value pairs, and extracting these pairs without user input or review.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual data entry is used to extract data from documents, then data can be entered into a database, but the process is tedious, slow, difficult to scale, and resource-intensive
Solution Approach 1:
The patent replaces manual mechanical data entry with an automated computer-based system that uses optical character recognition (OCR) and extraction rules to automatically extract data from documents and populate databases, eliminating the need for human operators to manually type data
Solution Approach 2:
The system enables documents to self-extract data through automated workflows where the computer-based tool independently processes documents, applies extraction rules, and populates databases without requiring human intervention at each step
2Measurement precision
If extraction rules are created for each document layout, then text can be converted to structured data, but a new set of rules must be created for each different layout
Solution Approach 1:
The patent creates a universal extraction system where a single set of extraction rules can handle multiple document layouts through the use of templates and configurable parameters, allowing the same rule set to adapt to different document formats without requiring separate rules for each layout
Solution Approach 2:
The extraction rules are made dynamic and configurable, allowing the system to adapt to different document layouts by adjusting parameters and templates rather than requiring fixed, layout-specific rules, enabling the same rule set to flexibly handle varying document structures
3Productivity
If machine learning models are trained on document content and structure, then predictions can be made for key-value pairs, but the models fail with documents that have dissimilar content or structure
Solution Approach 1:
The system performs preliminary processing of documents through OCR and region proposal generation before applying machine learning models, creating a standardized intermediate representation that helps the models generalize better across different document types by focusing on structural patterns rather than content variations
Data Source
AI summary
One embodiment of the present invention sets forth a technique for extracting data from a document. The technique includes determining a first set of features associated with the document, wherein the first set of features comprises a set of region proposals that bound one or more portions of text within the document. The technique also includes applying a first machine learning model to the first set of features to generate a set of predictions associated with one or more key-value pairs included in the document. The technique further includes extracting the one or more key-value pairs from the document based on the set of predictions.