2D Character Grid Layout for Semantic Document Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current document recognition methods fail to capture the spatial and visual structure of structured documents, which is crucial for understanding their semantic content, especially in formats like invoices, presentations, and blogs, as they rely solely on serialized text.

Innovation Solution

A computer-implemented method that utilizes a character grid representation to preserve the spatial structure of documents, converting them into a sparse 2D grid of characters using index values, enabling faster and more accurate processing with machine learning algorithms like convolutional neural networks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If current document recognition methods operate on serialized text, then processing is simpler and faster, but the spatial and visual structure of the document is lost

Engineering Contradiction:
Improveprocessing speedVSAvoidspatial structure
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The patent transforms the document representation from one-dimensional serialized text to a two-dimensional character grid that preserves spatial relationships. Each character is positioned according to its original location in the document, maintaining row and column information. This dimensional transformation allows the system to retain both the simplicity of structured data processing and the richness of spatial information.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent introduces a character grid as an intermediary representation between the original document image and the final semantic extraction. This grid serves as a bridge that converts visual spatial information into a structured format that machine learning algorithms can process while preserving positional relationships. The grid acts as a mediator that translates between visual and structural domains.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If document layout and positioning are preserved, then semantic understanding improves, but processing complexity increases

Engineering Contradiction:
Improvesemantic understanding accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the document into individual character units and organizes them into a grid structure. Each character is treated as a discrete element with specific positional attributes (row, column). This segmentation approach breaks down the complex task of understanding document semantics into manageable units while preserving their spatial relationships through the grid organization.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter representation from simple linear sequence indices to two-dimensional grid coordinates (row, column positions). This parameter transformation enables the system to capture spatial relationships without significantly increasing computational complexity, as the grid structure provides an organized framework for processing positional information.

Inventive Principle:
Principle #35Parameter changes

3Loss of information

If two-dimensional character grid is used, then spatial relationships are preserved, but data processing requires more sophisticated algorithms

Engineering Contradiction:
Improvespatial relationship preservationVSAvoidalgorithm sophistication
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent applies dimensional change by organizing characters in a 2D grid rather than a 1D sequence. This grid structure naturally preserves spatial relationships through row and column indices, allowing the system to maintain positional information while using standard matrix-based processing techniques that are well-suited for 2D data structures.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent creates a simplified copy of the document in the form of a character grid that captures essential spatial information without requiring processing of the full-resolution image. This copied representation maintains the necessary positional relationships while being much more efficient for algorithmic processing than the original visual document.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3570208B1Two-dimensional document processing
Publication Date: 2026.04.29 SAP SE
  • EP3570208B1 patent drawingFigure 1
  • EP3570208B1 patent drawingFigure 2A
  • EP3570208B1 patent drawingFigure 2B

AI summary

Disclosed herein are system, method, and computer program product embodiments for processing a document. In an embodiment, a document processing system may receive a document. The document processing system may perform optical character recognition to obtain character information and positioning information for the characters. The document processing system may generate a down-sampled two-dimensional character grid for the document. The document processing system may apply a convolutional neural network to the character grid to obtain semantic meaning for the document. The convolutional neural network may produce a segmentation mask and bounding boxes to correspond to the document.