Knowledge Question Answering Method For Patients With COPD Based On Knowledge Graph

Chronic obstructive pulmonary disease(Chronic Obstructive Pulmonary Disease,COPD)referred to as chronic obstructive pulmonary disease is a common and frequently-occurring disease,with a high case fatality rate,ranking third in the world as the cause of death.It is easy to diagnose whether or not COPD is diagnosed by functional examination,but the treatment level of patients diagnosed with COPD cannot keep up,because the medical resources of COPD are unevenly distributed,and even the treatment plans proposed by doctors in some areas are inconsistent with the standardized treatment in the COPD diagnosis and treatment guidelines.Hospital doctors have many patients,and patients have long time to see a doctor.Therefore,a question and answer about the treatment of COPD patients is needed to provide patients with help in treatment and reduce the burden on doctors.Therefore,for the treatment of patients diagnosed with COPD,this article proposes a question answering over knowledge base that uses the knowledge map of COPD as the knowledge source,uses natural language processing technology to convert user questions into available Cypher language queries,and returns the results to the user.The main research work and innovations of this article are as follows:(1)Because the research object is the treatment of patients diagnosed with COPD,according to the manual of "Diagnosis and Treatment of Chronic Obstructive Pulmonary Disease",expert doctors’ treatment opinions and the treatment plan in the electronic medical records of patients with COPD are the treatment standards.(2)By summarizing and analyzing the terms,concepts and relationships in the treatment standard,the conceptual model of the COPD knowledge graph is designed.(3)Take the COPD related data involved in the treatment standard as an entity set,fill in the entities according to the designed conceptual model,and generate a COPD knowledge map.(4)Use crawler technology to obtain all the original data sets involved in the treatment standard.(5)Through the named entity recognition algorithm based on BiLSTM-CRF,the COPD related entities in the original data set are identified to form the entity set of COPD knowledge map.(6)Knowledge quiz for COPD treatment is realized through two steps of question preprocessing and answer generation.The two-way maximum matching algorithm is used to segment the question sentence,and the type of question is judged by keywords.Use the LTP-parser tool to obtain the grammatical structure of the question,convert it into a question triple,and fill it into a Cypher query template that matches the question type to realize the question is converted into a knowledge graph query language and execute the statement.Get answers to questions.(7)Identify the question named entity to get the inconsistency between the entity and the knowledge graph,and use the similarity between the word vector and the character string to determine and solve it.