Research And Implementation Of Chinese Medical Automatic Question Answering System Based On Knowledge Graph

With the vigorous development of digital information technology,there are more and more ways for people to obtain medical and health information,but they lack professional knowledge to screen and discriminate.Starting from this need,this paper studies how to build a high-quality knowledge graph and build an automatic question answering system based on Chinese medical knowledge graph to help people obtain professional medical and health knowledge conveniently.Firstly,it explores how to build a high-quality knowledge map in the Chinese medical field.The whole process is roughly divided into data acquisition,knowledge definition and extraction,and knowledge storage.The first step is to use crawler technology to collect different forms of medical data on different medical consultation websites,and then define the three components of the knowledge map to be constructed: medical entities,relationships,and attributes,and finally in the collected raw data Extract these three parts and store them in the Neo4 j graph database.The second is how to design a solution that can efficiently complete the question answering task for a variety of questions in the question answering system.The question answering task process is divided into two parts,namely question parsing and classification and question answer data retrieval:1.Problem analysis and classification are divided into entity recognition,problem classification and intention detection.In order to achieve high accuracy in the recognition of medical entities,the strategy of entity dictionary plus synonym dictionary is adopted,and then the problems are classified through the grammar analysis results.Summarize three types of problems,and then propose an intent detection method that combines rule matching and existing deep learning methods to achieve intent detection for medical problems.achieved good results.2.The purpose of answer data retrieval is to return the data queried according to user questions from the knowledge graph,and organize these data into popular natural language and return them to users.The entity identified in the problem analysis is used as the root node of the knowledge map query,and the detected intent is used as the basis for judging which data associated with the entity needs to be returned.After obtaining the returned data,fill in the corresponding reply template and return it to the user Just complete the complete answering process for one question.Finally,the knowledge graph is combined with the automatic question-and-answer technology to realize a Chinese medical automatic answer system based on the knowledge graph,and finally access through the browser to realize the automatic question-and-answer to the user’s medical questions,and when the system cannot identify the user’s problem,it can Give users some guidance.By testing the system,it is proved that the Chinese medical automatic question answering system based on knowledge graph implemented in this paper can effectively answer users’ questions and meet the needs of the system.