Document Item Value Extraction Using Surroundings Context Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data management systems face challenges in reliably extracting item values from documents of a similar type due to differences in layout, particularly when the number of items or layout structure changes, leading to incorrect extraction of item values from corresponding regions in different documents.
Innovation Solution
An information processing apparatus that utilizes a processor to extract character strings from a second document by matching surrounding information acquired from the first document, using error correction and surroundings information to determine the correct item value extraction region, even when the layout differs between documents of the same type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If character string extraction is performed from a fixed region in documents of identical type, then extraction speed is improved, but extraction accuracy deteriorates when layout changes occur
Solution Approach 1:
The system performs preliminary action by acquiring and storing surroundings information from the first document before processing the second document. This pre-acquired context is then used to guide extraction in the second document, allowing the system to adapt to layout changes while maintaining efficient processing.
Solution Approach 2:
The surroundings information acts as an intermediary between the fixed extraction region and the variable content. By using this intermediate context (acquired from the first document), the system can accurately identify item values in the second document even when layouts differ, resolving the contradiction between speed and accuracy.
2Device complexity
If extraction region is fixed based on first document layout, then processing complexity is reduced, but extraction reliability worsens when item positions shift
Solution Approach 1:
The system creates a copy of the surroundings information from the first document and uses this copied context to guide extraction in the second document. This copying approach maintains simplicity while improving reliability, as the extracted item value is verified against the copied contextual information.
3Extent of automation
If pattern matching is used for item value extraction, then automation is improved, but accuracy deteriorates when character strings appear at undesired positions
Solution Approach 1:
The system implements feedback by using the surroundings information (acquired from the first document) to verify and correct extraction results in the second document. This feedback mechanism ensures that automated extraction maintains high accuracy by comparing against contextual patterns established from the first document.
Data Source
AI summary
An information processing apparatus includes a processor. The processor is configured to, in response to acquisition of a read image of a second document whose type is identical to a type of a first document, extract a character string included in, in the read image of the second document, a region where information about surroundings that matches first-document surroundings information is acquired, the information about surroundings satisfying a matching condition for determining matching with the first-document surroundings information that is acquired from an image around an item value acquisition region including an item value to be extracted in a read image of the first document, the character string being extracted as an item value to be extracted in the read image of the second document.