Question Answering System Based On Diabetes Knowledge Graph Research

Diabetes is a common disease.The question and answer system based on diabetes related medical knowledge will guide the diabetes patients to scientifically treat and alleviate diabetes,and guide diabetes patients to strengthen health management.In this thesis,the knowledge of diabetes treatment and the advantages of computer are combined to build diabetes knowledge Graph and realize the question answering system based on diabetes knowledge Graph.The main work of the system is named entity recognition,relationship extraction,knowledge Graph and the construction and implementation of question answering system.Named entity recognition and relationship extraction are important links in constructing knowledge Graph,and their performance will directly affect the quality of knowledge Graph.In order to improve the accuracy of relation extraction on diabetes data sets,this thesis proposes a Bilayer Spatiotemporal Graph Convolution Neural Network(Bi Sp GCN)for relation extraction.The model learns the temporal information of sentences based on Bi-directional Long Short-Term Memory(Bi LSTM),Then,the dependence tree is used as the input of Attention Guided Graph Convolutional Networks for Relation Extraction(AGGCN)to learn spatial information.Then,the Conditional Random Fields(CRF)is used to roughly label the temporal coding information of Bi LSTM,and the spatial coding information of AGGCN is deeply labeled to realize relationship extraction after feature fusion.The experimental results of this model on Chinese diabetes data set and Chinese character relationship data set show that compared with the existing AGGCN model,Bi Sp GCN model can fully learn the spatial and temporal information of sentences,achieve good results on Chinese diabetes data set,and have good performance for relationship extraction.At the same time,it can also directly obtain entity relationship triples and use them to construct knowledge Graph.The question answering system based on the diabetes knowledge graph can receive diabetes-related questions raised by users,and obtain the question category through algorithm analysis.Then get the corresponding answer through the graph database query statement.Finally,it is displayed on the web page in a visual way,thereby realizing the function of human-computer interaction.The process of building a diabetes knowledge map is: first,through named entity recognition and relationship extraction,entities and relationships are obtained from the Chinese diabetes dataset to form entity-relation triples,and then the Cypher statement that comes with the Neo4 j database is used to build a knowledge map.It is stored in the graph database in the form of a graph,and finally the knowledge graph is visualized through Neo4 j.This thesis focuses on the question answering system based on the knowledge Graph of diabetes.The QA system can help patients with assisted treatment and help users understand the knowledge of diabetes in order to prevent the onset of diabetes or alleviate the disease.