•  
  •  
 

Scientific Information Research

Keywords

Chinese medicine case， knowledge graph， knowledge graph complementation， link prediction， ERDBN model

Abstract

[Purpose/ significance]To propose a link prediction model for completing the knowledge graph of Chinese medicine cases, aiming to fill in the missing information in the knowledge graph,reveal the potential unknown associa‐ tions,and promote the construction of a more comprehensive and detailed knowledge graph of Chinese medicine cases. [Method/ process]Medical cases in the field of TCM asthma were collected and integrated to construct a knowledge graph of TCM asthma diagnosis and treatment. By introducing Dropout regularization and Batch Normalization tech‐ niques into the RotatE model, an improved ERDBN model is proposed and applied to the task of refining the asthma diagnosis and treatment knowledge graph. [Result/conclusion]The ERDBN model performance is first tested on four knowledge graph public datasets FB15K,FB15K-237, WN18 and WN18RR to verify the effectiveness of the model in the knowledge graph complementation domain, and the experimental results of the metrics Hits@1 are improved by 1.0%, 2.0%, 0.5% and 2.3%, respectively. In the task of complementing the knowledge graph of asthma diagnosis and treatment based on the construction of TCM medical cases, the ERDBN model successfully predicted the tongue infor‐ mation related to the lung solid and kidney deficiency syndrome and the lung and kidney both deficiency syndrome, which verified the validity and practicality of the model. This study provides a strong support for completing the knowl‐ edge graph of TCM medical cases and promoting the construction and application of TCM diagnosis and treatment knowledge graph.

Digital Object Identifier (DOI)

10.19809/j.cnki.kjqbyj.2025.03.010

Download

Share

COinS