Research On Knowledge Graph Question Answering Method And Its Application In Automatic Diagnosis

With the rapid development of Internet technology,people’s daily life has been inseparable from the network.More and more questions have changed from book query to network query,and automatic question answering is a new trend of human-computer interaction.Analyzing the semantic information in natural language questions put forward by users and obtaining accurate answers according to their real intention will become a new form of the next generation search engine.The traditional automatic question answering is to obtain answers by retrieving unstructured data such as web pages,search engines,knowledge encyclopedia and question answering communities.With the application of deep learning methods such as knowledge map and neural network in the field of natural language processing,the automatic question answering method based on knowledge map has attracted extensive attention.Question answering based on knowledge map aims to analyze the semantics of natural language questions and return accurate answers from knowledge map.At present,Automatic Question Answering Based on knowledge map has achieved initial results for simple problems with single relationship and problems without temporal reasoning relationship,while problems in reality are more complex due to temporal constraints or multiple relationships.When the questions raised by users are complex,especially when they contain temporal relationships,due to the weak reasoning ability of the model and insensitive to the temporal constraints in the questions,they often cannot get accurate answers.It is also a challenge faced by the automatic question answering system,which makes the question answering system difficult to implement in many fields,Especially in medical knowledge consultation and medical service,the quality is poor.Aiming at the above problems,this paper proposes a method of using external knowledge and reasoning enhancement,and applies the method to the automatic inquiry prototype system to further verify the effectiveness of the method.The main work and innovations of this paper are as follows:(1)For the problem of insensitive timing constraints,we integrate external knowledge into the automatic question answering model through the attention mechanism,and enhance the sensitivity of the model to timing and other information by paying attention to the parts that are not paid attention to by the general model,such as timing information,so as to further improve the accuracy of problem prediction.(2)Aiming at the problem of weak reasoning ability of the model,we design a reasoning enhancement network to analyze the context information and integrate it into the reasoning process,so as to improve the accuracy of relations and entities in the reasoning process.(3)The method of integrating external knowledge and reasoning to enhance the network is applied to the actual automatic consultation system,which further proves the practicability of this method.