Dual Recognizer System for Reducing Text Input Workload
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data input systems require significant human workload for text recognition, as they rely solely on recognition probabilities to determine whether first or second processing is necessary, leading to inefficiencies in workload distribution.
Innovation Solution
An information processing apparatus that utilizes both a first and second recognizer to acquire recognition probabilities, allowing for the determination of which processing to execute based on these probabilities, thereby reducing human workload by employing first processing with a lower workload when possible.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If only a first recognizer is used to determine processing type based on recognition probability, then the system complexity is low, but the human workload is excessive
Solution Approach 1:
The patent combines multiple recognizers (first recognizer and second recognizer) to work together in a unified system. The first recognizer performs initial text recognition while the second recognizer verifies the results, merging their capabilities to reduce human workload without excessive complexity increase.
Solution Approach 2:
The second recognizer acts as an intermediary between the first recognizer and the final output. It verifies recognition results and determines whether manual input is needed, mediating the process to reduce unnecessary human intervention while maintaining system reliability.
2Reliability
If manual verification is performed for all low probability recognition results, then the recognition accuracy is high, but the processing time increases
Solution Approach 1:
The patent applies different processing qualities to different recognition results based on their probability. High probability results undergo automatic verification with minimal human intervention, while low probability results receive full manual verification, creating local quality variations that optimize both accuracy and time efficiency.
Solution Approach 2:
Instead of performing full manual verification on all recognition results, the system performs partial verification only when necessary. The second recognizer identifies cases where manual input is truly needed, applying verification action partially rather than excessively to all inputs.
3Reliability
If the threshold for manual input is lowered, then the recognition accuracy improves, but the productivity decreases
Solution Approach 1:
The system dynamically adjusts the threshold for manual input based on recognition probability. Rather than using a fixed low threshold that would reduce productivity, the threshold is dynamically determined by the second recognizer's verification of the first recognizer's confidence level, optimizing both accuracy and efficiency.
Data Source
AI summary
An information processing apparatus includes a processor configured to execute first acquisition processing for acquiring a first recognition result and a first recognition probability on target data from a first recognizer, execute second acquisition processing for acquiring a second recognition probability for the first recognition result on the target data from a second recognizer, and execute control for determining which of first processing and second processing with a necessary human workload greater than in the first processing is to be executed for the first recognition result based on the first recognition probability and the second recognition probability.