Repeating Object Group Detection for Unstructured Document Editing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing document processing technologies struggle to identify and process repeating structure groups of objects in documents without structural information, leading to inefficient and manual editing processes, particularly in visually-rich documents like brochures and presentations.
Innovation Solution
A digital design system employs a heuristic-based method using a deep-learning page segmentation model and heuristics such as symmetry, alignment, and proximity to identify repeating structure groups of objects in documents, allowing for automated editing and propagation of changes across multiple instances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual editing methods are used for objects in documents without structural information, then editing can be performed on individual objects, but editing efficiency is low and manual intervention is required for each object
Solution Approach 1:
The patent merges multiple individual objects that share common attributes into a single grouped object or template. By combining repeating objects (such as multiple figures, tables, or text elements with identical formatting) into a unified structure, editing one instance automatically updates all others, thereby dramatically improving editing efficiency and eliminating repetitive manual interventions.
Solution Approach 2:
The patent creates universal templates that can represent multiple specific objects with similar characteristics. A single template definition serves multiple instances across the document, allowing one editing action to propagate universally to all matching objects, thus resolving the contradiction between individual object editability and overall editing efficiency.
2Adaptability or versatility
If documents are processed without structural information, then document formats like PDF can be widely used, but identifying and processing repeating structure groups becomes difficult
Solution Approach 1:
The patent replaces traditional mechanical/document-structure-based object identification methods with machine learning-based detection. By using trained models to automatically analyze visual patterns, spatial relationships, and attribute similarities, the system can identify repeating structure groups in unstructured documents without requiring predefined structural information, thus maintaining format versatility while overcoming the detection difficulty.
Solution Approach 2:
The patent transforms the detection approach by changing from structure-based parameters to attribute-based parameters. Instead of relying on document structural metadata that doesn't exist in formats like PDF, the system analyzes visual attributes (position, size, content similarity, formatting) to identify repeating objects, enabling structure group detection in previously unstructured documents.
Data Source
AI summary
Embodiments are disclosed for a process of detecting and processing repeating structure groups of objects in a document using a digital design system. The method may include obtaining, by a page segmentation model, object information for a plurality of objects in a document. The disclosed systems and methods further comprise determining, using the object information, a plurality of object clusters based on distances between the plurality of objects. A repeating structure group of objects can then be identified using the plurality of object clusters. The disclosed systems and methods further comprise providing information indicating the repeating structure group of objects in the document