Construction And Reasoning Of Pediatric Knowledge Graph Based On Neo4j

With the wide application of medical information technology,the reasoning model based on the pediatric knowledge graph has become a new type of knowledge reasoning model for pediatric disease diagnosis.The traditional method of constructing a knowledge graph based on triples cannot express probabilistic pediatric knowledge;the traditional method based on the deep learning inference engine is difficult to deal with the problem of input imbalance caused by the common input of different numbers of symptoms;the deep learning inference engine-based Traditional methods can only output one result.Therefore,this thesis proposes a method of constructing a paediatric knowledge graph with quaternions,visualized with Neo4 j software,and establishing an Adapt-Trans E-NBC model as an inference engine.First of all,the traditional method cannot display the probabilistic pediatric knowledge,which causes the pediatric knowledge map to not reflect the characteristics of the data set accurately,clearly or intuitively.Based on the original RDF triples,this thesis forms a data structure of quadruples by adding the corresponding relationship probability to the relationship of some triples.Then,based on part of the original data in the form of RDF triples,part of the data in the form of newly formed quaternions,and Neo4 j construction software,a paediatric knowledge map is jointly constructed and visualized,so that the constructed paediatric knowledge map can express probabilistic data.Paediatric knowledge,and can serve the reasoning engine based on the paediatric knowledge graph.Secondly,according to the existing methods,inputting different numbers of symptoms in the dataset into the same model makes it difficult for the inference engine to deal with the problem of input imbalance,which leads to the problem of low accuracy in multiple diseases.In this thesis,an Adapt-Trans E-NBC inference model based on the paediatric knowledge graph is established.The Adapt-Trans E-NBC inference model first converts the translation model Trans E into the corresponding Trans E inference engine according to its own score function and relevant rules of link prediction,and then converts the translation model Trans E into the corresponding Trans E inference engine.Originally,the inference results were determined by a single Trans E inference engine,and the inference results were jointly determined by the Trans E inference engine and the Naive Bayes classifier.At the same time,the Adapt-Trans E-NBC inference model also introduces an adaptive mechanism to improve the performance of the model according to the characteristics of the input data.Through the established adaptive mechanism,each model processes a pediatric dataset with an approximate number of inputs,and makes each model It is no longer a simple combination,but training appropriate collocation parameters according to the data set to reduce the impact of input imbalance,thereby improving the inference accuracy of the inference engine.Finally,due to the characteristics of the black box,the traditional method can only provide one result,that is,the final result,but cannot provide the second most possible reasoning result for reference.The Adapt-Trans E-NBC model calculates the inference results of the inference engine numerically,so that it can output more than one inference result for reference.In this thesis,"Ask a Doctor" and data sets related to pediatrics on Open KG are used for experiments.It is proved through experiments that the paediatric knowledge graph is constructed by quadruples,and the method of visualization with Neo4 j software enables the probability value to be displayed on the edge of the knowledge graph,which can better reflect the data and serve the inference engine.Compared with Text-CNN and Bi-LSTM/BLSTM,the Adapt-Trans E-NBC model has improved in the precision,recall and F1 value of children’s various diseases reasoning,so the proposed method can be used for answering questions on health network.Consulting and improving the accuracy of medical diagnosis provide a more reliable basis and can provide more than one result for easy reference.