Research On Construction Method Of Medical Imaging Knowledge Graph And Implementation

Knowledge graph is an efficient knowledge expression model proposed by Google,which can map things in the real world,and provides an effective means to organize,manage and apply things in the real world.In the medical field of Internet,knowledge graph is the main means of medical consultation system and medical intelligent question-and-answer system.However,at present,these Internet medical consultation systems are mainly aimed at disease-centered question-and-answer systems,and imaging examination,as an important means for finding and locating diseases in modern medicine,plays an important role in modern clinical medicine.There are many ways of medical imaging examination,such as x-ray,CT,MR,DR,PET-CT,PET-MR,each of which has its own characteristics and applicable scope.Therefore,it is of great significance to construct medical knowledge map centered on imaging examination.Aiming at the key technology of Named Entity Recognition in the construction of knowledge graph,aiming at the problem that the traditional Named Entity Recognition based on character sequence ignores the word information in the text and has poor recognition effect on rare words and unregistered words,this paper combines the characteristics of named entity recognition in Chinese medical field,and incorporates lexicon information,font glyph features and morphological formation features of characters into the input layer of the traditional model.The effectiveness of the method is verified by experiments.Aiming at the key technology of knowledge fusion in the construction of knowledge graph,this paper proposes a knowledge fusion algorithm based on rule and similarity fusion,which uses rule matching,semantic similarity calculation and semantic web similarity calculation to fuse entity,and proves the effectiveness of the method through experiments.This paper first introduces the related background knowledge and theoretical technology of knowledge graph construction,then analyzes the requirements of the system to be constructed in detail,and expounds the core functions and performance indicators of the system.Then,it focuses on the algorithm of Chinese medical named entity recognition,and solves the key problems of this paper.Then,according to the requirement analysis and outline design of the system,the detailed design and implementation of the system are described.Finally,a complete functional test and performance test are carried out for the application subsystem.