Research And Implementation Of Question Answering System Based On Medical Knowledge Graph Of Common Diseases

With the development of society and the improvement of people’s living standards,people’s living habits have undergone tremendous changes in recent years,and their own health requirements are getting higher and higher.The early symptoms of some diseases make it difficult for the general public to detect,and when suffering from diseases,many people have a large knowledge blind spot about their physical state,often delaying the best treatment time,and need a way to understand these diseases without queuing for registration.The development of Internet technology has made some medical information can be stored in various websites and databases in the form of electronic data,but the accuracy of electronic data has become a problem in application.In recent years,artificial intelligence has become the mainstream of today’s era,and knowledge graph,as an important branch of artificial intelligence,has been rapidly popularized and applied in all walks of life in recent years,providing the possibility for Internet companies to use professional knowledge to build a question and answer system based on knowledge graph in related directions.Combined with the current social demand for medical knowledge of common diseases,the development of a question and answer system based on the medical knowledge map of common diseases will facilitate people’s daily life by taking the electronic data of common diseases as the source of information of the knowledge graph and text classification as the technical support of the question and answer system.The development of question and answer system based on medical knowledge map of common diseases still faces the following problems: how to understand user intention more accurately in question and answer module.How to improve the accuracy of relationship and entity extraction in the intelligent construction of common medical knowledge map.To solve the above problems,this paper takes the deep learning correlation algorithm as the research focus construction system,and conducts in-depth research on the text classification and text classification correlation algorithm of the question answering module related to entity relationship extraction in the process of map construction.The main work and innovation of this paper are as follows:(1)The extraction accuracy of various professional knowledge relationships in the medical field.In this paper,BERT is used as the data preprocessing model,combined with Bi-LSTM+CRF for entity relationship extraction.Compared with traditional entity extraction algorithm,neural network correlation algorithm is more accurate for entity extraction.The reason is that neural network has the function of sensing the context,and can understand the meaning of the word in the sentence according to the context,and can extract relevant professional knowledge more accurately.(2)Aiming at the accuracy of classification intention identification of Chinese text in question and answer module.In this paper,attention mechanism combined with BERT+Text CNN fusion model is used to automatically screen medical information about common diseases proposed by users.The traditional classification of each sentence can only be fixed,but the deep learning correlation algorithm can be used for context classification.(3)Realize the question and answer system based on knowledge graph.The system takes the information obtained from 39 Health network as the data source and constructs the knowledge map automatically by extracting the data after entity relationship.The knowledge graph is used as the data source of the question answering system,and the relevant information of the graph database is queried after text classification to give the question answer.