Binary Feature Dictionary Construction for Semantic Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semantic similarity models constructed using neural networks face challenges in introducing new granularity information, requiring a method to extract and utilize binary features effectively for improved accuracy.
Innovation Solution
A method and apparatus for constructing a binary feature dictionary by extracting binary features from a corpus, calculating a preset statistic for each feature, and selecting a predetermined number of features based on this statistic to form the dictionary, which can then be used to introduce new features into the training data of the semantic similarity model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If new granularity information is introduced to improve semantic similarity model accuracy, then model accuracy improves, but the complexity of extracting and processing new features increases
Solution Approach 1:
The patent segments the feature extraction process into distinct stages: extracting binary features from corpus, calculating statistics for each feature, and selecting top features to form a dictionary. This segmentation makes the complex extraction process manageable and systematic, resolving the contradiction between improving accuracy through new features and managing extraction complexity.
Solution Approach 2:
The patent performs preliminary actions by pre-processing the corpus to extract binary features and pre-calculating their statistics before model training. This preliminary feature dictionary construction simplifies the subsequent model training process and reduces the complexity of feature processing during training, while still enabling the model to benefit from new granularity information.
2Loss of information
If binary features are extracted and selected to form a feature dictionary, then new granularity information is introduced, but the time and computational resources required for feature processing increase
Solution Approach 1:
The patent extracts only the most relevant binary features from the corpus by calculating statistics and selecting top features based on predetermined criteria. This extraction approach ensures that new granularity information is introduced without processing all possible features, thereby reducing the time and computational resources required while maintaining information completeness.
Solution Approach 2:
The patent changes parameters by selecting a predetermined number of top binary features based on statistical measures rather than using all extracted features. This parameter-based selection optimizes the balance between information completeness and processing efficiency, reducing feature processing time while maintaining the quality of new granularity information introduced to the model.
Data Source
AI summary
Disclosed are a method and an apparatus for constructing a binary feature dictionary. The method may include: extracting binary features from a corpus; calculating a preset statistic of each binary feature; and selecting a preset number of binary features in sequence according to the preset statistic to constitute the binary feature dictionary.