计算机技术与发展2025，Vol.35Issue(3)：140-147,8.DOI:10.20165/j.cnki.ISSN1673-629X.2024.0341
融合知识图谱与图卷积神经网络中医证型分类模型
A Classification Model of TCM Syndrome Based on Integration of Knowledge Graph and Graph Convolutional Network
摘要
Abstract
The classification of traditional Chinese medicine(TCM)syndrome is an integral component of the TCM diagnostic system.The accuracy of syndrome classification affects the effectiveness of TCM diagnosis and treatment,so how to improve the accuracy of syndrome classification has always been an important topic in TCM research.Leveraging artificial intelligence technologies to explore TCM syndrome classification models aims to enhance the performance of syndrome classification,thereby laying the groundwork for other related applications in TCM.We employ a hybrid model—KGRGCN,which is capable of efficiently capturing the deep features between data nodes.Graph Convolutional Networks(GCN)are effective for handling graph-structured data,and by incorporating residual structure,the model's expressive power and training stability are significantly enhanced.Integrating syndrome-related knowledge graphs helps the model merge syndrome embeddings,thereby enhancing the relationships between symptoms and syndromes.To further enhance model performance,we propose a strategy that incorporates status element weights as a bridge,integrating multi-layer information repre-sentations.Additionally,a Multi-Layer Perceptron(MLP)is employed for syndrome classification.This approach aims to leverage the combined strength of diverse information layers and the classification power of MLP to achieve more accurate and robust results in syndrome analysis.The experimental results indicate that the KGRGCN model proposed performs excellently in syndrome classification tasks.Specifically,the model achieved an accuracy of 75.43%,a precision of 74.93%,a recall of 76.91%,and an F1-score of 75.91%.These results demonstrate that the model outperforms several popular classification methods,including SVM,TextCNN,and Random Forests,in terms of classification performance.关键词
图卷积神经网络/证型分类/知识图谱/证素/残差结构Key words
graph convolutional network/syndrome classification/knowledge graph/status element/residual structure分类
计算机与自动化引用本文复制引用
董云春,晏峻峰..融合知识图谱与图卷积神经网络中医证型分类模型[J].计算机技术与发展,2025,35(3):140-147,8.基金项目
湖南省教育科学研究重点项目(23A0312) （23A0312）
湖南中医药大学研究生创新课题项目(2024CX085) （2024CX085）
