Traditional information retrieval cannot fully understand the semantics of the question,and the query results are often confusing and messy which need to be screened artificially.Especially,in the field of medical consultation,the complexity and rigor in medical knowledge make the study to get accurate and concise answers of domain question become a research direction of concern and promising for researchers.In addition,with the wide application of information technology in the medical field,people search for related problems on the Internet more and more frequently.In the medical field,there is an urgent need for a consultation and question answering information system,such as "family doctor",which can offer medical assistance and solve basic medical problems.On the one hand,the goal of automatic question answering research is to provide accurate and concise answers.On the other hand,Ontology organizes and manages knowledge effectively,and can understand the semantics of knowledge with an semantic reasoning ability.Therefore,taking the Ontology of medical field as an example,this paper proposes an ontology-based automatic question-answering model with semantic reasoning,which strives to use the advantages of Ontology in knowledge management and reasoning to achieve cheap,efficient and accurate medical services such as medical question-answering and diagnostic advice.The main work is as follows:(1)Analyzing and integrating medical data in medical field,giving the method of extracting knowledge in medical field,and constructing a medical domain Ontology.Based on the Ontology,the process of how to implement the automatic question answering based on Ontology is analyzed.(2)Aiming at the deficiency of Static Framework which can not be automatically updated when used to understand the semantic of question.Based on CFN framework,giving a model of using Ontology to automatically generate QFN framework,which provides a new understanding strategy for question analysis.Then to extract the corresponding answers,the algorithm for generating SPARQL automatically are also given,which is based on RDF triple query graph and deep learning.The method transform the process of Seq2 seq into classification,which avoids the challenge of network parameters and structure adaptive adjustment and large training corpus brought by the complexity of seq2 seq model.(3)Designing experiments by constructing an automatic question answering system to validate and evaluate the effectiveness of the method.Experiments show that the Ontology-based automatic question answering system in the medical field can improve the accuracy of the answer to question and has a good performance in question reasoning,which can be widely applied and promoted. |