The Construction And Application Of Knowledge Graph Based On The Ancient Books Of Traditional Chinese Medicine

Traditional Chinese Medicine is the treasure of ancient Chinese medicine and the crystallization of the thoughts,culture and wisdom of doctors of all ages.Ancient books of traditional Chinese medicine are the main carrier of knowledge for the development of traditional Chinese medicine,recording thousands of years of medical knowledge and practical experience of the Chinese nation,such as Huangdi Neijing,treatise on febrile diseases,synopsis of the golden chamber and other ancient books of traditional Chinese medicine,which carry the core knowledge of basic theories and Clinical Prescriptions of traditional Chinese medicine,and have great medical research and clinical value after thousands of years of clinical verification and development.In the era of artificial intelligence and big data technology,the mining and utilization of ancient Chinese medicine literature knowledge is one of the important basic tasks for the inheritance and innovation of traditional Chinese medicine,but there are still significant challenges as follows:(1)The capacity of ancient Chinese medicine books is huge,and they are recorded in the form of books,the data is mainly in the form of unstructured text,the manual processing of ancient Chinese books data such as named entities and so on It takes time and effort to extract.(2)Ancient books and documents are recorded in classical Chinese.The words used in classical Chinese are concise,which is quite different from modern literature in terms of vocabulary and semantics.The standard data set used for artificial intelligence analysis is particularly lacking.Therefore,it provides a big obstacle for computer methods to automatically extract ancient books and documents.(3)For the purpose of clinical diagnosis and treatment assistance,there are also big problems in how to quickly query and apply the relationship knowledge of ancient Chinese medicine books.In view of the above problems,this paper mainly studies from the following three aspects.1)The research of named entity recognition of ancient books of traditional Chinese medicine based on deep neural network.In this paper,we mark and audit the named entities of ancient books and documents by manual method,and form a standard corpus with 1179408 samples.The corpus contains 11 named entity types and 1895210 entities,which provides an important data base for the research of named entity extraction methods for ancient books of traditional Chinese medicine.We build word embedding model by Word2 Vec,ELMo and BERT as input,and the task of named entity recognition is carried out by BILSTM + CRF model.In the final experimental results,BERT + BILSTM + CRF model achieved the best experimental results,with the precision of 83.07% and F1 value of 83.25%.Among them,the precision of Word2Vec+ BILSTM + CRF model with relatively poor results is also 80.16%,and the F1 value is80.34%.In general,good experimental results are obtained in ancient book entity extraction.2)Research on entity relationship extraction of ancient Chinese medicine books based on deep neural network.First of all,we annotate and verify the entity relationship of ancient books and documents data manually and form a standard corpus.Finally,we get 90705 relationship data from 662 ancient books and documents.Based on this relation extraction standard corpus,we use the pipeline method of BERT + BILSTM +CRF and PCNN + ATT to extract the direct relation.Among them,PCNN + ATT method is used to extract the relationship,and 63.25% accuracy and 63.57% F1 value are obtained.In contrast,the pipeline method based on BERT + BILSTM + CRF for named entity recognition and relationship extraction gets 61.43% precision and 61.99%F1 value.The experiment also verifies that the attention mechanism based method achieves relatively good results in relation extraction.3)Research on the construction of knowledge graph of ancient Chinese books and case analysis.The neo4 j graph database is used to build the knowledge graph database of ancient Chinese books.The relationship data of artificial annotation and intelligent extraction audit are stored in the knowledge graph platform,forming a total of 119380 entities and 164028 relationships of ancient Chinese books knowledge base.Furthermore,the front-end visualization platform based on neo4 j provides convenient technical support for the query and utilization of the knowledge graph.