Heading Extraction via Line Direction Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sentence analysis systems face challenges in accurately segmenting documents into sections due to non-uniformity or errors in heading assignment, leading to incorrect segmentation, especially when headings are described at the same position in the line direction.
Innovation Solution
A sentence structure analysis system that processes documents by obtaining headings based on their position related to the line direction, using both regular and over-extracted patterns to identify and segment sections accurately, thereby enhancing the accuracy of heading extraction and document segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If headings are extracted based on style and expression detection, then the system can identify potential section titles, but the accuracy of heading extraction decreases when there is non-uniformity or error in heading assignment
Solution Approach 1:
The patent changes the extraction parameter from style/expression detection to positional detection. Specifically, it uses the line direction position (horizontal position) of text as the key parameter to identify headings. Headings are extracted based on their consistent positioning at the beginning of lines, which is a structural parameter rather than a stylistic one. This parameter change makes the extraction more reliable even when heading styles vary.
Solution Approach 2:
The patent introduces a new dimension for heading extraction by using the spatial position (line direction) as the primary criterion. Instead of analyzing the content or style of text, the system examines the positional dimension - specifically, whether text appears at the beginning of lines in a consistent pattern. This dimensional shift from content analysis to position analysis resolves the contradiction.
2Adaptability or versatility
If the system processes documents with non-uniform heading assignment, then it can handle diverse document formats, but the document cannot be correctly segmented for each section
Solution Approach 1:
The patent changes from content-based parameters (style, expression) to structural parameters (position, line direction) for heading identification. This parameter change enables the system to handle diverse document formats adaptively while maintaining segmentation precision, because positional structure is more consistent across different document styles than content features.
Solution Approach 2:
The patent segments the document processing into distinct steps: first extracting headings based on positional criteria, then using these extracted headings to segment the document into sections. This segmentation approach allows the system to handle diverse formats in the first step while achieving accurate segmentation in the second step, resolving the contradiction between adaptability and precision.
Data Source
AI summary
A sentence structure analysis system is provided. A line direction is a direction in which characters are arranged in a document. Headings are obtained from the document based on a position related to the line direction. The document is segmented into a first block and a second block. The first block includes one or more sentences from a first heading to a row preceding a second heading. The second block includes one or more sentences from the second heading to a row preceding a third heading.