Design And Implementation Of Text-based Medical Knowledge Graph

Medical knowledge graph is one of the research hotspots of artificial intelligence technology in the field of medicine.The current medical knowledge information on the Internet is generally of poor quality.With the application of knowledge graphs in the medical field,it has become possible to popularize medical knowledge among the general population.The medical knowledge graph not only provides people with intuitive and accurate medical knowledge,but also relieves the shortage of high-quality medical resources in China to a certain extent.This paper proposes a design and implementation of text-based medical knowledge graph.Able to construct a knowledge graph from the massive medical literature according to the process proposed in this article.In this important part of knowledge extraction,the medical literature is first identified for named entities.On this basis,a rule-based method is used to extract sentence sets that contain candidate triples,and then relationship extraction is performed to extract triples.In the task of named entity recognition and relationship extraction,this paper builds the FB＿T＿CRF model and FBO＿T＿ATT model based on the transformer structure.On the Chinese diabetes data set of Ali Tianchi,the F1 scores reached 86.59 and 88.26 respectively,which fully shows that the model proposed in this paper has strong predictive ability.This paper implements a text-based medical knowledge graph system,which can extract medical knowledge from massive medical literature and provide visual query of knowledge.This article shows the design process of the system,including role design,server deployment and web programming,as well as some screenshots of the business process of the system during operation,which proves that the system has high practical value and analyzes the shortcomings of the current system Improve direction before and after.