Construction Of Knowledge Graph Of Lingshu Sutra Based On Joint Extraction Method

As an important external therapy in Traditional Chinese Medicine,acupuncture has attracted wide attention at home and abroad because of its unique therapeutic effect,and has become the most widely used traditional and alternative medicine in the world.As the foundation work of acupuncture,Lingshu Sutra plays an important role in the establishment of the theoretical system of acupuncture and moxibustion,and it is also an important ideological source of scientific research and innovation of modern acupuncture and moxibustion.In particular,on the basis of the inheritance and development of a variety of acupuncture and moxibustion ancient books,recording the theoretical knowledge,practical experience and classical cases summarized by famous doctors of all ages,it is a great treasure house of acupuncture scholars.How to sort out and integrate the theoretical knowledge recorded in Lingshu Sutra and more acupuncture ancient books,and explore the potential laws of acupuncture diagnosis and treatment,so as to provide support for the inheritance and innovation of modern acupuncture and moxibustion? It has become one of the important problems to be solved urgently for acupuncture scholars in the digital age.With the further combination of medicine and information technology,a large number of multi-source heterogeneous data such as electronic medical records,medical records and books are used for secondary mining to assist clinical decision-making and provide support for modern intelligent diagnosis and treatment services.The ancient books of acupuncture and moxibustion are mostly stored in the form of unstructured text,which is not convenient to record the integration,induction and mining of knowledge,resulting in a large number of idle resources and waste;second,there is a lack of norms and standards in various text processing tasks.A large number of noise affects the effectiveness of mining rules,while the direct use of general models can not effectively capture the professional knowledge in the text.Therefore,how to achieve accurate and efficient extraction of professional knowledge from acupuncture texts,and then explore the potential rules of diagnosis and treatment,in order to better serve clinical diagnosis and treatment,inheritance and innovation,has become one of the research focuses and difficulties of acupuncture scholars.The application of knowledge graph in the field of acupuncture provides a feasible solution to the above problems.Aiming at the problems of lack of corpus,difficulty in term recognition and low quality of knowledge acquisition in acupuncture and moxibustion ancient books,this paper constructs a word segmentation hybrid model which combines Traditional Chinese Medicine Ancient Books Pre-Training Model,Bi-directional Long Short Term Memory and Conditional Random Field to achieve more accurate professional term recognition,and proposes a span-based joint extraction model to effectively extract professional knowledge from texts.The knowledge graph based on joint extraction is constructed and analyzed visually in order to further mine the potential rules.The main contents of this paper are as follows:(1)Based on the BERT Pre-Training Model,the CmabBERT model is obtained by secondary training.The BERT model is based on the Encoder layer of the Transformer model,and captures the data features through the attention mechanism.After a large number of TCM ancient book mask prediction tasks are retrained,we can further learn and master the context features in the context of TCM.The analysis of the mask prediction effect before and after the second training shows that the CmabBERT model has a better ability to predict the context of TCM texts.At the same time,it is applied to the word segmentation task of acupuncture ancient books to improve the performance of the model.(2)Construct CmabBERT+BiLSTM+CRF fusion model for word segmentation.In this model,the dynamic word vector representation of acupuncture text is output by CmabBERT model,and the text features are extracted by BiLSTM layer.Finally,the word segmentation sequence is tagged by CRF layer to identify the terms in the corpus.Through the analysis of the comparative results,the recognition ability of the model is more in line with the needs of professional vocabulary recognition.(3)A joint extraction model based on span is constructed for knowledge extraction.Based on the BERT model,the model generates a set of entity candidates in an arbitrary span and then determines the category to which the entity belongs,and then determines the corresponding relationship through the arrangement and combination of entities to achieve knowledge extraction from the text.The test consequence indicates that the model can implement the good recognition of most entities and relationships in acupuncture ancient books,and lays a good foundation for the construction of knowledge graph.(4)Triplet data storage and knowledge graph visualization are realized based on Neo4 j graph database.Neo4 j stores data in the form of unstructured points and edges,which can quickly respond to complex queries and intuitively display the data in the way of network graph.The triple knowledge extracted from the ancient books of Lingshu Sutra is stored in Neo4 j to construct a knowledge graph to realize the systematic and multi-dimensional display and analysis of fragmentary and flat knowledge in ancient books.Combined with deep learning and knowledge graph technology,this study comprehensively combs and analyzes a large number of theoretical knowledge,practical experience and clinical cases recorded in acupuncture ancient books,excavates the potential rules of diagnosis and treatment,and constructs the knowledge graph of Lingshu Sutra based on ontology framework,which provides a more accurate strategy and more efficient service for the intelligent diagnosis and treatment of traditional Chinese medicine.The results of this study can provide reference and reference for the innovative research of acupuncture scholars,and contribute to the inheritance,innovation and intelligent development of acupuncture.