Sentence Classification Apparatus Parameter Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sentence classification methods fail to appropriately classify sentences due to unadjusted parameters during the vector conversion process, leading to ineffective classification of clinical trial conditions and failure coping methods described in free text.
Innovation Solution
A sentence classification apparatus and method that adjusts parameters by obtaining case sentences associated with effect values, creating case values, calculating correlation coefficients between case and effect values, and selecting optimal parameters for accurate classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If parameters are left unadjusted during sentence-to-vector conversion, then the process is simple and fast, but the classification accuracy deteriorates
Solution Approach 1:
The patent systematically varies parameters including dimensionality (50, 100, 150, 200, 250, 300), window size (3, 5, 7, 9, 11, 13), and sampling rate (0.1, 0.2, 0.3, 0.4, 0.5, 0.6) to identify optimal configurations for sentence vector conversion. This exhaustive parameter search resolves the contradiction by finding settings that achieve high classification accuracy without requiring complex manual adjustment during operation.
Solution Approach 2:
The patent performs preliminary parameter optimization by evaluating all parameter combinations in advance using correlation coefficients with effect values. The optimal parameters determined through this preliminary action are then fixed and applied to subsequent classification tasks, eliminating the need for complex real-time parameter adjustment while maintaining high accuracy.
2Measurement precision
If parameters are optimized through exhaustive search, then classification accuracy improves, but computational time increases
Solution Approach 1:
The patent employs an exhaustive parameter search that goes beyond what is strictly necessary, testing all combinations of dimensionality (6 values), window size (6 values), and sampling rate (6 values). This excessive action ensures finding the global optimum for parameter settings, which then can be applied repeatedly without re-optimization, reducing overall time loss in practical applications.
Solution Approach 2:
The system performs self-optimization by automatically evaluating parameter combinations and selecting those with highest correlation coefficients with effect values. This self-service approach eliminates the need for manual parameter tuning and automates the optimization process, making the time investment worthwhile through sustained high accuracy performance.
3Reliability
If more parameters with different values are tested, then the likelihood of finding optimal settings increases, but the complexity of the selection process increases
Solution Approach 1:
The patent implements feedback mechanisms by calculating correlation coefficients between sentence vectors and effect values for each parameter combination. This feedback quantifies the performance of each parameter setting, enabling systematic selection of optimal parameters. The feedback loop ensures reliable classification by continuously evaluating parameter effectiveness against the target effect values.
Data Source
AI summary
A sentence classification apparatus, a sentence classification method, and a sentence classification program that can appropriately classify sentences by adjusting parameters are provided. The sentence classification apparatus includes: a case sentence obtention unit for obtaining plural case sentences which are associated with effect values that are values obtained by evaluating effects; a case value creation unit for creating case values obtained by numerizing the case sentences for each of parameters with different values; a correlation coefficient calculation unit for calculating a correlation coefficient between the case values and the effect values for each of the values of the parameters; and a parameter selection unit for selecting a parameter among the parameters with different values on the basis of the correlation coefficient.