Text Classification Model Using Graph Convolutional Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current text classification methods do not effectively utilize hierarchical relationships between categories, leading to suboptimal accuracy in predicting text classification categories.
Innovation Solution
A service model incorporating an encoding network, a fusion layer, and a graph convolutional network is used to process text and category hierarchical relationships, generating semantic and category expression vectors that are fused to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hierarchical relationships between categories are not utilized, then the classification method is simple, but the text classification accuracy is suboptimal
Solution Approach 1:
The patent introduces a hierarchical dimension to the category structure by constructing a category hierarchical relationship network where categories are organized across multiple levels (e.g., first-level categories like 'finance' containing second-level categories like 'finance management' containing third-level categories like 'stocks'). This dimensional transformation allows the model to capture semantic relationships at different granularities, thereby improving classification accuracy without excessive complexity increase
Solution Approach 2:
The patent introduces a graph convolutional network as an intermediary component that processes the category hierarchical relationship network. This intermediary structure aggregates semantic information from neighboring categories and propagates hierarchical features through the network, enabling the model to effectively utilize category relationships while maintaining a manageable architectural complexity through modular design
2Measurement precision
If hierarchical category relationships are incorporated, then classification accuracy improves, but the processing complexity increases
Solution Approach 1:
The patent replaces traditional mechanical classification approaches with a graph neural network-based semantic processing system. Instead of using rule-based or shallow matching methods, the system employs graph convolutional networks to automatically learn and propagate semantic features through the hierarchical category structure, reducing the difficulty of processing complex hierarchical relationships through automated feature learning
Data Source
AI summary
Implementations of the present specification provide a method and an apparatus for training a service model, and a method and an apparatus for determining a text classification category. During specific implementation, on the one hand, text is processed by using an encoding network to obtain a corresponding semantic vector; on the other hand, a relationship network is established for classification categories based on a hierarchical relationship, and the relationship network is processed by using a graph convolutional network, to fuse information of nodes to obtain category expression vectors. Then, the semantic vector of the text is fused with the category expression vectors to determine a prediction result of a classification category. In a phase of training a service model, the prediction result can be compared with a sample label to determine a loss and adjust model parameters. In a phase of determining a text classification category by using a trained service model, the corresponding classification category can be determined based on the prediction result. This implementation can improve text classification accuracy.