Research And Application Of Semantic Analysis In Medical Questions And Answers System

With the rise of Internet diversification,artificial intelligence promotes the process of "Internet+medical",facilitates the storage and retrieval of data in the medical field,and accumulates a large amount of medical text data.Massive medical texts contain valuable medical knowledge.The study of knowledge map can better construct the knowledge base of medical field,and facilitate people to mine and construct semantic relations in massive medical text data.The process of constructing knowledge map is also the establishment of people's cognition of the objective world.In this paper,a knowledge graph is constructed for the vertical field of medical treatment,and the constructed medical domain knowledge graph is used as a medical knowledge base,and a medical question-and-answer system is designed.The main difference between the medical question and answer system of knowledge graph and the traditional medical question and answer system is the storage form of data.Compared with the traditional medical question-and-answer system,its storage form is no longer a question-and-answer pair,but a multi-relation semantic network graph composed of entity nodes and relationships in the medical field,and each entity node has its attribute content,so that the knowledge base in the medical field can have a good description ability.The question-and-answer system of medical knowledge base designed in this paper is mainly divided into the following three aspects:(a)crawling of basic medical data.Due to the lack of publicly available data sets of Chinese medical domain knowledge,scapy framework technology was used to climb medical knowledge in vertical medical domain.(b)construction of medical knowledge base.The medical data was cleaned and the entity relation edge was created in the Neo4j graph database to realize the visualization of the medical knowledge base.(c)question analysis of medical question answering system.Aiming at the medical information named entity recognition and the classification of medical questions,this paper analyzes the semantic meaning of the medical questions raised by users.This system question parsing module through the deep learning model Bert-CRF(Bidirectional Encoder Representations from Transformers-CRF)and TextCNN essay this classification model medical questions semantic parser is designed.In this paper,the bert-crf model was adopted.On the CCKS2018 electronic medical record data set,the accuracy of named entity recognition of the bert-crf model was higher than that of the Bert model alone.