Text Semantic Similarity With Structural–Plane Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating text semantic similarity rely solely on vector representations, which fail to capture the nuanced differences in meaning between texts with similar character compositions, leading to inaccurate similarity calculations.
Innovation Solution
A method that combines text structural features, optimized through scaling and standardization, with text plane features using a multi-layer perceptron, and utilizes a support vector regression model with a kernel function to calculate text similarity, enhancing the accuracy of similarity assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If text similarity is calculated solely based on vector representations, then the calculation process is simple, but the accuracy of semantic similarity is poor
Solution Approach 1:
The patent segments text analysis into multiple feature dimensions: text plane features (character-level patterns) and text structural features (syntax trees, dependency relationships). This segmentation allows comprehensive semantic analysis while maintaining modular processing, resolving the contradiction between accuracy and complexity by organizing complex features into manageable segments.
Solution Approach 2:
The patent combines multiple feature types (plane features and structural features) into a composite feature representation. By integrating character-level patterns with syntactic and semantic structures, the system achieves high-precision semantic similarity measurement, effectively resolving the contradiction between measurement precision and feature extraction complexity.
2Measurement precision
If text structural features are extracted and combined with plane features, then the accuracy of similarity calculation is improved, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary feature extraction and scaling before similarity calculation. By pre-processing text into plane features and structural features, and applying feature scaling in advance, the system reduces computational burden during the actual similarity computation, effectively managing the trade-off between accuracy and computational power.
Solution Approach 2:
The patent applies feature scaling to transform structural feature values into standardized ranges. This parameter transformation optimizes the feature representation for computational efficiency while preserving semantic information, allowing high-accuracy similarity calculation with reduced computational resources.
3Reliability
If feature scaling and standardization are applied to text structural features, then the optimization of feature representation is achieved, but the processing time increases
Solution Approach 1:
The patent applies feature scaling and standardization to normalize structural feature values. These parameter transformations improve feature representation quality and reliability by ensuring consistent value ranges, while the efficient implementation minimizes the time penalty, effectively balancing reliability and processing time.
Data Source
AI summary
The present disclosure relates to the technical field of text analysis, and discloses a method for calculating text semantic similarity, including: acquiring a text pair to be compared, and respectively extracting a text structural feature of each text in the text pair to be compared; performing feature scaling and feature standardization processing on each text structural feature to obtain an optimized text structural feature of each text; acquiring a text plane feature of each text in the text pair to be compared, and combining the optimized text structural feature of each text with a corresponding text plane feature to obtain a structure-plane feature of each text; and learning the structure-plane feature of each text to obtain a text similarity of the text pair to be compared, scoring the text similarity to obtain a text similarity score. The present disclosure can improve the accuracy of text similarity calculation.