Text Cluster Ranking for Visually Rich Document Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to efficiently process and rank important information in visually-rich documents, such as vaccination certificates, due to the lack of effective methods for combining visual layout analysis with semantic understanding, leading to increased computational resources and time in downstream applications.
Innovation Solution
A hybrid method that combines visual property ranking based on layout structure with semantic context ranking, using neural networks to identify and rank text clusters, reducing computational load by selecting only the most important clusters for downstream processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all text clusters in visually-rich documents are processed by downstream applications, then complete information extraction is achieved, but computational resources and processing time increase significantly
Solution Approach 1:
The patent applies preliminary action by performing visual layout analysis and semantic understanding ranking before downstream processing. The system pre-identifies and ranks text clusters based on their visual properties and semantic relevance, so that only the most important clusters are subsequently processed by downstream applications. This preliminary filtering step reduces the volume of data requiring full processing while ensuring that critical information is not lost.
Solution Approach 2:
The patent extracts only the essential text clusters from visually-rich documents based on visual layout structure and semantic context. By taking out and selecting only the most relevant text clusters for downstream processing, the system avoids processing redundant or less important information, thereby reducing computational resources and time while maintaining information extraction completeness for critical data.
2Productivity
If visual layout analysis and semantic understanding are combined to rank text clusters, then processing efficiency is improved, but system complexity increases
Solution Approach 1:
The patent merges visual layout analysis and semantic understanding into a unified ranking system. By combining these two analysis approaches, the system achieves more accurate text cluster ranking that considers both visual prominence and semantic relevance. This merging of multiple analysis methods improves processing efficiency by providing a comprehensive ranking that guides downstream applications more effectively, despite the increased system complexity.
Solution Approach 2:
The patent creates a multi-functional ranking system that performs both visual layout analysis and semantic understanding through a unified framework. This universal system handles multiple types of document analysis (visual properties, semantic context, ranking) within a single integrated approach, improving overall processing efficiency while managing complexity through consolidation rather than separate independent systems.
3Speed
If only visual properties are used to rank text clusters, then processing speed is maintained, but information accuracy decreases
Solution Approach 1:
The patent applies local quality by differentiating the ranking criteria based on the specific characteristics of text clusters. While visual properties provide a baseline ranking that maintains processing speed, the system locally enhances the ranking by incorporating semantic understanding for text clusters where it matters most. This localized addition of semantic analysis improves information accuracy for critical clusters without requiring semantic analysis for every single cluster, thus preserving overall processing speed.
Data Source
AI summary
Computer-implemented methods of automatically identifying, ranking, and processing information obtained from a document, and computerized systems and computer program products related thereto. The method involves identifying text clusters and identifying a visual layout structure of at least one part of the document, and ranking the text clusters according to visual properties of the text cluster. The method further involves identifying a semantic context of the identified text clusters and ranking the text clusters according to a similarity of the identified semantic context in relation to a given semantic context, to obtain a semantic context ranking, creating a total ranking of the text clusters based on a combination of a pair of rankings, and selecting text cluster(s) according to its position in the total ranking and providing the selected text cluster(s) to at least one downstream application.