Dynamic Keyword Extraction Across Document Pages
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing key-value extraction technologies fail to extract target information when it is not recorded near the keyword, as they rely on a predetermined range based on the keyword's position, leading to unsuccessful extraction in documents like contracts or complaints where the target information is referenced elsewhere.
Innovation Solution
An information processing device with a processor that determines the presence of a referral character string indicating the target information and extracts it from a specified reference, even if it's not within the initially defined range based on the keyword's position, using character recognition and pattern matching to locate the target information across pages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If key-value extraction technology is used to extract target information from a predetermined range based on keyword position, then extraction speed is improved, but extraction accuracy deteriorates when target information is not recorded near the keyword
Solution Approach 1:
The patent makes the extraction range dynamic by adjusting it based on document structure analysis. Instead of using a fixed predetermined range, the system dynamically determines the appropriate search scope by analyzing document headings, sections, and structural elements, allowing the extraction range to adapt to different document formats and layouts.
Solution Approach 2:
The patent adds a new dimension to the extraction process by incorporating document structure analysis (headings, sections, pages) as an additional layer beyond simple positional range search. This multi-dimensional approach combines structural understanding with keyword-based extraction to locate target information more accurately across the entire document.
2Device complexity
If the extraction range is limited to a predetermined area based on keyword position, then device complexity is reduced, but adaptability deteriorates when dealing with various document formats
Solution Approach 1:
The patent segments the document into structural elements (headings, sections, pages, paragraphs) and processes each segment independently. By dividing the document structure into manageable parts, the system can analyze and navigate different document formats systematically without requiring a completely different approach for each format type.
Solution Approach 2:
The patent creates a universal extraction framework that can handle multiple document formats through a single integrated system. The document structure analysis module serves multiple functions: identifying sections, determining search ranges, locating target information, and adapting to different layouts, making the system versatile across various document types without increasing complexity.
Data Source
AI summary
An information processing device includes a processor programmed to: determine whether a character string indicating a reference to target information is recorded in a range within a document, the target information corresponding to a keyword in the document, the range being determined based on a position of the keyword in the document; and extract the target information upon a determination that the character string is recorded in the range.