LLM Document Attribute Extraction Across Varying Layouts

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing text extraction tools are unable to reliably search and extract specific pieces of information from unstructured documents due to their varying and complex layouts, requiring advanced techniques to handle diverse formats like GIF, JPEG, PNG, and TIFF without human intervention.

Innovation Solution

Employing artificial intelligence (AI) large language models (LLMs) to convert unstructured documents into text format, allowing users to select page-level and document-level information, and create prompts for the LLM to accurately extract requested data without human intervention, converting the output into structured formats like JSON.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional search tools are used on unstructured documents, then the search process is simple, but the text cannot be searched because it is stored as images

Engineering Contradiction:

Improvesearch capabilityVSAvoidtext searchability

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent replaces manual text extraction methods with optical character recognition (OCR) technology to automatically convert images of text into machine-readable text formats, enabling search capabilities on unstructured documents while maintaining ease of operation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary processing layer that converts unstructured document images into structured text data through OCR and natural language processing, allowing conventional search tools to effectively query unstructured documents

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If unstructured documents are scanned and saved for reference, then document preservation is achieved, but searchability is lost

Engineering Contradiction:

Improvedocument preservationVSAvoidsearchability

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent creates a digital copy of the text content from scanned documents through OCR technology, preserving the original scanned document while generating a searchable text representation that maintains the information without losing searchability

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent applies preliminary text extraction and structuring processing to scanned documents during the preservation process, converting images to searchable text formats before storage, so that searchability is established in advance rather than requiring later conversion

Inventive Principle:

Principle #10Preliminary action

3Device complexity

If traditional text extraction tools are used on documents with varying formats and complex layouts, then the process is simple, but extraction accuracy deteriorates

Engineering Contradiction:

Improveextraction process simplicityVSAvoidextraction accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent employs dynamic processing that adapts to varying document formats and layouts by using machine learning models that automatically adjust extraction parameters based on the specific characteristics of each document, maintaining both process simplicity and high extraction accuracy

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes processing parameters dynamically based on document characteristics, using OCR technology with adjustable recognition parameters and natural language processing to optimize extraction accuracy for different formats, layouts, and languages while maintaining a unified simple interface

Inventive Principle:

Principle #35Parameter changes

4Measurement precision

If human intervention is used for information extraction, then extraction accuracy is high, but the process requires manual involvement

Engineering Contradiction:

Improveextraction accuracyVSAvoidmanual involvement

Core Design Contradiction:

Measurement precisionVSExtent of automation

Solution Approach 1:

The patent implements self-service automated extraction using OCR technology and natural language processing models that automatically identify and extract relevant information from unstructured documents without human intervention, achieving both high accuracy and full automation

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent incorporates feedback mechanisms where the system learns from extraction results and continuously improves accuracy through machine learning, automatically adjusting processing parameters based on performance metrics to maintain high extraction accuracy without manual involvement

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20260064741A1Method and system for extracting information from documents with varying formats

Publication Date: 2026.03.05 SCHLUMBERGER TECH CORP

AI summary

Certain aspects of the disclosure provide a method for extracting attributes from documents with varying formats, layouts and complexities. The method displays a user interface (UI) that enables a user to obtain an unstructured document from a knowledge base. The method converts the unstructured document into a text document using a text recognition. The method obtains, as output from a large language model (LLM), an extracted page attribute from the text document. The extracted page attribute contains a first type of information recorded in text on a single page of the text document. The extracted document attribute contains a second type of information recorded in text on more than one page of the text document. The method obtains, as output from the LLM, an extracted document attribute from the text document. The extracted page attribute and the extracted document attribute are displayed in the UI.