Research And Implementation Of Intelligent Question Answering Based On Deep Learning In Medical Knowledge Graph

With the gradual advancement of China’s medical informatization construction,improving the level of medical and health services and perfecting and improving the medical service system are the urgent problems that need to be solved at present.Therefore,China has put forward the policies related to the construction of intelligent medical care to improve the national health awareness and provide convenient intelligent medical services.However,because the information construction in the field of medical health is still not perfect,at the same time,with the improvement of people’s quality of life,traditional hospitals can no longer meet people’s medical needs,and people are eager to realize the query of certain medical general knowledge or online medical assistance consultation through more convenient ways.Knowledge graph-based medical intelligence Q&A is expected to solve this problem.By deep semantic parsing of the question,medical intelligent Q&A based on knowledge graph can understand the intention of the question and provide accurate answer feedback to users with the help of existing expertise graph,which meets people’s higher requirements for medical services.In order to respond to the national policies in the medical field and to meet the medical needs of the public,this paper starts from the implementation of intelligent Q&A in the general medical field,constructs a knowledge map about the medical field and implements an automatic Q&A based on deep learning on this basis.The main research of this paper includes the following:(1)Using Python web crawler technology,we crawl structured or semi-structured data from relevant medical websites,integrate them,build a knowledge graph based on medical field,store them using Neo4 j graph database,and visualize the knowledge graph.(2)To realize intelligent Q&A based on medical knowledge graph,named entity recognition and intention recognition are used to achieve the semantic parsing task of medical question sentences,firstly,a named entity recognition model based on BERT+Bi LSTM+CRF is designed and implemented to recognize the main entity of medical question sentences,secondly,a classification dataset about medical question sentences is constructed,and a pre-trained model BERT+Text CNN text classification model is designed based on the pre-trained model BERT+CRF to recognize the main entity of medical question sentences.Text CNN text classification model is designed to classify the medical question based on the pre-training model BERT+Text CNN,so as to determine the final intention of the question,and the results of recognition and classification are transformed into standard query statements for answer retrieval in the knowledge graph.(3)Building a medical intelligent question and answer system,using the Django web development framework and integrating the above related contents,a medical intelligent question and answer system based on deep learning is built to achieve accurate answers to users’ medical question and answer system,which satisfies users’ needs for medical question and answer system.By integrating the above research contents,this paper completes a complete deep learning-based medical knowledge graph Q&A system,which can achieve accurate answers to users’ medical questions and answers.The research content of this paper provides some reference experience for future researchers in the direction of building medical question and answer systems,and has some practical significance and application value.