Multi-Scale Convolutional Network for Text Similarity Checking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods using convolutional neural networks for text similarity checks are limited by their fixed scale convolution, resulting in low flexibility and accuracy.
Innovation Solution
The proposed method employs a pre-trained convolutional neural network with multi-scale convolutional layers, lexing texts to generate word vector sequences, combining eigenvector sequences, and analyzing them using a similarity prediction model to determine text similarity, enhancing flexibility and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed-scale convolutional neural network is used for text similarity checking, then the network structure is simple, but the flexibility and accuracy of text checking are reduced
Solution Approach 1:
The patent segments the text checking process into multiple convolution operations with different window lengths (e.g., 3, 4, 5, 6, 7, 8, 9, 10). Each convolution operation extracts features at a different scale, allowing the system to capture both local and global text characteristics simultaneously, thereby improving flexibility without excessive complexity
Solution Approach 2:
The patent introduces a multi-dimensional approach by applying convolutions with varying window lengths across the same text input. This adds a dimensional aspect of scale variability to the feature extraction process, enabling the system to analyze text at multiple granularities and improve adaptability
2Measurement precision
If a fixed-scale convolutional neural network is used, then the computational complexity is low, but the accuracy of text similarity detection is reduced
Solution Approach 1:
The text is segmented and analyzed at multiple scale levels through parallel convolution operations with different window lengths. This multi-scale segmentation allows the system to detect both fine-grained local patterns and coarse-grained global structures, significantly improving detection accuracy
Solution Approach 2:
The patent combines multiple convolution results from different window lengths into a composite feature representation. This composite approach integrates information from various scales, creating a more comprehensive and accurate text similarity detection capability
Data Source
AI summary
This disclosure discloses an artificial intelligence based method and apparatus for checking a text. An embodiment of the method comprises: lexing a first to-be-checked text and a second to-be-checked text respectively, determining word vectors of the lexed words to generate a first word vector sequence and a second word vector sequence; inputting the first word vector sequence and the second word vector sequence respectively into a pre-trained convolutional neural network containing at least one multi-scale convolutional layer, identifying vector sequences in a plurality of vector sequences outputted by a last multi-scale convolutional layer as eigenvector sequences, to obtain eigenvector sequence groups respectively corresponding to the texts; combining eigenvector sequences in each eigenvector sequence group to generate a combined eigenvector sequence; and analyzing the generated combined eigenvector sequences to determine whether the first text and the second text pass a similarity check. The embodiment improves the flexibility in checking a text.