Research On Question Answering Of TCM Syndrome Elements Based On Knowledge Graph

Traditional Chinese medicine(TCM)is a thousand-year-old medical tradition in China that embodies the wisdom and experience of numerous predecessors.However,with the passage of time and social progress,the inheritance and development of TCM face a series of challenges and difficulties.Among them,the cultivation of TCM talents is a key aspect.TCM practitioners not only need to grasp abundant theoretical knowledge of TCM but also possess clinical practice experience and skills.The academic thoughts and experiences of renowned senior TCM experts need to be effectively inherited.Their unique treatment techniques and methods are valuable treasures for the development of TCM.However,there are still issues with the loss of some unique treatment techniques and methods,which not only impact the inheritance and development of TCM but also constrain the construction of TCM talent pool.In order to make up for this deficiency,the progress in the construction of hospital information systems provides new opportunities for the inheritance and development of TCM.Various TCM electronic medical records,classical TCM literature,and online publications contain abundant knowledge of TCM and information on disease diagnosis.Extracting high-quality information from these TCM databases and fully utilizing it is crucial for improving the quality of clinical decision-making in TCM.However,in actual TCM clinical diagnosis,the standardization of TCM syndrome differentiation results is challenging.Diagnoses often vary from person to person or from disease to disease,and there is a lack of unified standardization in the textual representation of syndromes.The diagnostic thinking of "four diagnostic methods combined with syndrome differentiation" in TCM leads to more complexity in the extraction and utilization of knowledge,posing further challenges for the informatization of TCM diagnosis.To address this issue,this paper proposes an automatic question-answering system based on a TCM knowledge graph.The system breaks down syndromes to obtain disease nature and disease location,transforming the multi-classification task of syndromes in TCM diagnosis into a multi-label classification task of syndrome elements.It also constructs a TCM knowledge graph.The results of multi-label classification serve as answers for syndrome element questions,while also enabling general TCM knowledge-based question answering.The main work of this paper is summarized as follows.(1)Construction of the Traditional Chinese Medicine Knowledge Graph(TCMKG):Design and construct the TCM knowledge ontology based on data characteristics,gather medical data from TCM electronic medical records,ancient TCM literature,and various vertical websites.Then,use knowledge representation-based entity alignment methods to achieve knowledge fusion and complete the construction of the TCMKG.(2)Design and implementation of the multi-label classification algorithm incorporating knowledge graph information: Utilize knowledge representation learning methods to obtain representation vectors for entities and relationships in the constructed TCMKG.A fusion method combining characters and entities is designed to add the representation information of the knowledge graph to the word vector information of the pre-trained BERT model,thereby enhancing the multi-label classification performance of the model.Experimental results demonstrate that incorporating implicit knowledge graph representation learning information can improve the accuracy of the model.(3)Design and implementation of the automatic question-answering system based on the Traditional Chinese Medicine Knowledge Graph: The question-answering design consists of two parts.The first part is the syndrome element question-answering driven by the multi-label classification model.The second part includes recommending similar syndromes and general knowledge-based question-answering.This system can assist doctors in clinical diagnosis and provide scientific references for patients to understand TCM knowledge and choose appropriate treatments.