Research On The Construction Of Medical Knowledge Graph QA System Based On Deep Learning

The QA system has been extensively researched in the field of NLP and has been applied to many specific areas such as customer service and financial information services.Knowledge-based entities and relationships in the open knowledge Graph continue to grow,providing a more complete information base for knowledge-based question-and-answer systems.However,open domain knowledge Graph have certain limitations for question-andanswer queries in specific fields,and lack relevant knowledge information in specific fields.At the same time,user queries mostly use natural language description problems,and there are differences between the data structures stored in the knowledge graph.Based on deep learning,it is constructed for the medical domain knowledge graph QA system,and is dedicated to the knowledge graph of the medical field and the construction of the question and answer system based on the graph.Firstly,the system uses crawler related technology to capture the raw data of medical websites,and transforms them into structured data storage through data processing and cleaning.By analyzing the data,defining the entities,relationships and attributes in the medical knowledge graph,and constructing medical knowledge graph.Then use the Word2 Vec network training word vector and the TextCNN problem classification algorithm to construct a high-precision problem classification module.In order to improve the accuracy,the system introduces the heuristic extended entity detection based on the bidirectional short-short time memory network and the AR-SMCNN relationship detection algorithm based on the similarity matrix matching,constructs the question and answer module,and conducts a comparative experiment on the benchmark data set against this module to verify the effect of question and answer.Finally,on the constructed medical knowledge graph,the QA system based on the template matching method and the deep learning method is used to compare the questions and answers and verify the effect.This paper uses the Flask lightweight development framework to integrate the data collection,data processing and storage,knowledge graph construction,QA query and other modules in the QA system,and provides a QA page to facilitate users to ask questions about medicines or diseases and provide disease data.The query and display page displays the association information of the disease,and the user can query the specific disease,provide the knowledge graph query and management page to facilitate the knowledge management of the knowledge graph.