Context-Aware Document Augmentation for Heterogeneous Artifacts

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing document processing technologies fail to capture the contextual relationships and interdependencies of heterogeneous document artifacts, leading to incomplete digital intelligibility and metadata extraction limitations.

Innovation Solution

A system utilizing OCR extraction and structured document format processing, combined with general and image-processing large language models, generates augmented documents by adding contextually aware natural language descriptions to document artifacts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If traditional document processing technologies are used, then document conversion to machine-readable format is achieved, but contextual relationships and interdependencies of heterogeneous document artifacts are not captured

Engineering Contradiction:
Improvecontextual relationshipsVSAvoidprocessing system complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The document processing system segments the document into multiple heterogeneous artifacts (tables, images, forms, text sections) and processes each artifact type through specialized models. This segmentation allows comprehensive capture of contextual relationships while maintaining manageable processing complexity through modular architecture.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system introduces an intermediary layer of natural language descriptions that mediates between the raw document artifacts and the final digital representation. This intermediary layer captures contextual relationships and interdependencies, transforming structured data into semantically enriched representations that preserve document context.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of information

If comprehensive natural language descriptions are generated for all document artifacts, then digital intelligibility is enhanced, but processing time increases

Engineering Contradiction:
Improvedigital intelligibilityVSAvoidprocessing time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

The system applies partial action by selectively generating natural language descriptions based on artifact type and complexity. Different artifact types are processed through specialized models with varying levels of detail, ensuring comprehensive coverage where needed while avoiding unnecessary processing time for simpler artifacts.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system changes processing parameters dynamically based on artifact characteristics. The level of natural language description generation is adjusted according to artifact type, size, and complexity, optimizing the balance between digital intelligibility enhancement and processing time consumption.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If multiple specialized models are used for different artifact types, then processing accuracy is improved, but system complexity increases

Engineering Contradiction:
Improveprocessing accuracyVSAvoidmodel architecture complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system implements universality through a unified document processing framework that handles multiple artifact types (tables, images, forms, text) through a common architecture. This multi-functional approach maintains high processing accuracy for each artifact type while reducing overall system complexity through shared components and standardized processing pipelines.

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

Data Source

PatentEP4657308A1Context aware document augmentation and synthesis
Publication Date: 2025.12.03 INTUIT INC
  • EP4657308A1 patent drawingFigure 1
  • EP4657308A1 patent drawingFigure 2
  • EP4657308A1 patent drawingFigure 3

AI summary

A method includes obtaining a document structure from a data repository. The document structure includes multiple structured sections. A table is detected in a first structured section. A table representation of the table is processed by a general large language model (LLM) to generate a natural language description of the table. An image is detected in the first structured section. The image is processed by an image-processing LLM to generate a natural language description of the image. A form is detected in the first structured section. The form is processed by the general LLM to generate a natural language description of the form. The natural language descriptions of the table, image and form are inserted into the first structured section to obtain a modified first structured section. A modified document structure including the modified first structured section is outputted.