Research On Question Answer System For Chronic Disease Medical Care Based On Knowledge Graph

In recent years,despite the rapid development of society and the continuous improvement of living standards,the fast-paced life has also led to the frequent occurrence of chronic diseases in people’s daily life,which has seriously affected people’s normal life.How to conveniently and quickly find useful medical knowledge for our daily prevention and treatment of chronic diseases becomes particularly important.The amount of relevant knowledge queried from search engines through traditional methods is very large and scattered,and the efficiency of obtaining the required knowledge from it is very low.The question answering system based on knowledge graph can solve this problem very well.This paper mainly studies the construction of chronic disease knowledge graph and the realization of question answering system.The chronic disease knowledge graph is constructed through the chronic disease medical knowledge data obtained from the network,semantic analysis is performed through named entity recognition and intent recognition,and the answers to user questions are obtained from the knowledge graph,so that users can query disease-related knowledge.The main work of this paper is as follows:（1）Through the analysis of the data in the relevant websites through the crawler technology,the original data was obtained by the crawler script.After cleaning these data,the clean data needed to build the knowledge graph is obtained.The data is then classified and the analysis of the data defines categories of entities,relationships and attributes.Finally,the classified data is stored in the Neo4 j database in the form of triples,and a chronic disease knowledge map is constructed.（2）In the question answering system,the method of semantic parsing is used to analyze the question sentence,which mainly includes the task of named entity recognition and the task of intent recognition.In the named entity recognition task,based on the traditional Bi LSTMCRF,a Ro BERTa-Bi GRU-CRF model is proposed to convert the question into a semantic vector by introducing the improved model Ro BERTa of the pre-trained model BERT.Different from Bi LSTM,Bi GRU reduces model training parameters and trains faster.The vector obtains contextual semantic information through Bi GRU and finally performs entity recognition on the question sentence through CRF.In the task of intent recognition,the Ro BERTa model is used to classify questions.Finally,by combining the database query statement corresponding to the intent recognition classification with the entity identified by the named entity,the answer is obtained from the chronic disease knowledge graph stored in the Neo4 j database.（3）On the basis of the above work,a chronic disease medical question answering system based on knowledge graph is realized.It mainly provides users with chronic disease medical question answering services and knowledge graph visualization.