Research On Construction Of Hypertension Knowledge Graph Based On Deep Learning

In recent years,with the rapid development of China’s economy,our living standard has greatly improved,but people work with heavier and heavier pressure,more overtime,more social intercourse and less physical exercise,unhealthy lifestyle leads to the rapid growth of hypertension.Hypertension has the characteristics of occult onset,long course and difficult to control.The death rate and disability rates of hypertension are high,which seriously affects the working ability and life quality of the patients.The high cost of hypertension treatment aggravates the economic burden of the patient’s family and social medical care.How to effectively prevent and control the spread of hypertension has become an important issue.Knowledge graph describes the concept in the objective world and the relationship between them.At present,Recommendation System,Intelligent Search,Knowledge Q & A System and other upper applications are relying on knowledge graph as the underlying service.Therefore,this paper is intended to study the framework of hypertension knowledge graph construction and the improvement in method.The main work is as follows:1.Named entity recognition is an important part of knowledge extraction,which mainly extracts meaningful words from text.And named entity is an important component of knowledge carrier,so the recognition result of named entity directly affects the effect of knowledge graph.In this paper,using domain knowledge graph to improve the precision of named entity recognition.This method applies knowledge graph to deep learning model,which improves the application scope of knowledge graph.2.There are two common ways to construct knowledge graph: top-down and bottom-up.The top-down construction method is used by high-quality data to extract ontology and pattern information manually or automatically,and then build knowledge graph.The bottom-up construction method is to use deep learning model to extract knowledge information on massive data,and then construct knowledge graph.The top-down method need domain experts,with the increase of data scale,the manual construction becomes more difficult.This paper presents a data-driven construction method,which reduces the participation in domain experts and the requirements of knowledge graph construction.3.Knowledge graph is an intelligent,efficient and accurate way of knowledge organization.However,the construction process of knowledge graph is very time-consuming and labor-consuming.Therefore,how to utilize the existing knowledge graph to accelerate the construction of new knowledge graph has chief research values.Transfer learning can solve the above problems.In view of this,based on the existing diabetes knowledge graph,using the migration learning method,accelerate the construction of hypertension knowledge graph.In this paper,we use the method of model based Transfer Learning to initialize the target domain model,and use the Instance based Transfer Learning to build auxiliary samples,when the labeled samples are less,so as to improve the model results.Experimental results show the effectiveness of the proposed method.