Document Retrieval Using Associating Keywords for Handwriting Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of character recognition for handwritten documents remains insufficient due to variations in handwriting styles, leading to defects and erroneous recognitions, which hinder effective keyword searches and retrieval of desired documents.
Innovation Solution
A document retrieval system that includes an input reception unit, a document acquisition unit, a keyword acquisition unit, a document search unit, and a search result output unit, utilizing an associating keyword database to associate authors' names with keywords and perform fuzzy searches, and optionally employing a prediction model learning unit to improve search accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If character recognition is performed on handwritten documents using current technology, then electronic files can be created for retrieval, but the recognition accuracy is insufficient due to handwriting variations
Solution Approach 1:
The patent introduces an intermediary correction process between OCR and keyword search. When a keyword search is performed, if no results are found, the system automatically generates corrected keywords by replacing potentially misrecognized characters with candidate corrections, then performs search again with these corrected keywords. This intermediary correction mechanism resolves the contradiction by maintaining adaptability to handwriting variations while improving recognition accuracy.
Solution Approach 2:
The system implements feedback by using search results to inform keyword correction. When the initial keyword search yields no results, the system feeds back to the keyword processing stage, generating corrected keywords based on character recognition confidence and candidate substitutions, then re-performs the search. This closed-loop feedback mechanism continuously improves search accuracy despite handwriting variations.
2Productivity
If keyword search is performed on digital document data with recognition errors, then search speed is maintained, but the hit rate decreases due to erroneous character recognition
Solution Approach 1:
The patent applies partial correction action by selectively correcting only those keywords that fail to return search results. Instead of correcting all keywords universally (which would reduce search speed), the system performs correction only when necessary (when initial search yields no results), thus maintaining search productivity while improving reliability for failed searches.
Solution Approach 2:
The system performs preliminary character recognition and stores confidence information before keyword search. This preliminary action allows the system to quickly identify potentially misrecognized characters and generate correction candidates without delaying the initial fast search, thus maintaining search speed while preparing for potential corrections to improve hit rate.
3Ease of manufacture
If traditional OCR is used to convert handwritten documents to text, then electronic storage is achieved, but character recognition accuracy remains insufficient
Solution Approach 1:
The system implements self-service by automatically correcting its own recognition errors during the keyword search process. When a search fails, the system autonomously generates corrected keywords by replacing low-confidence characters with candidate corrections and re-performs the search, without requiring manual intervention. This self-correcting mechanism maintains the ease of automated digitization while improving recognition accuracy.
Solution Approach 2:
The patent replaces the traditional mechanical OCR process with an intelligent, adaptive search system that incorporates automated error correction. Instead of relying solely on the mechanical accuracy of character recognition, the system substitutes a smarter search mechanism that can identify and correct recognition errors through automated keyword generation and iterative searching, thus improving accuracy without complicating the digitization process.
Data Source
AI summary
A document retrieval apparatus includes a processor which receives an input of a keyword, acquires an author's name and a document file from a digital document database which stores document files of text data obtained by performing a character recognition process with respect to document image data of handwritten documents, and names of authors who wrote the handwritten documents, references an associating keyword database which stores information associating the authors' names, keywords, and associating keywords, to acquire an associating keyword of the input keyword, from the received input keyword and the acquired author's name, searches the acquired document file, using the input keyword and the acquired associating keyword, and outputs a search result of the searching.