Design And Implementation Of Chinese Medical Question Answering System Based On Deep Learning

With the popularization of the Semantic Web,more and more structured data has appeared on the Internet.The stored data led by the RDF format also plays a huge role in a large number of open fields and specific fields,making ordinary users on the Internet The demand for this part of the data is also increasing.However,because ordinary users do not understand the characteristics of this structured knowledge,it is relatively difficult to directly access the structured knowledge.To solve this problem,many traditional knowledge-based question answering systems have been proposed.This question answering system is more than traditional search engines.Smart,can directly understand the search needs of ordinary users.Taking the Chinese medical field as the research background and taking deep learning as the technical basis,this paper proposes and implements a Chinese medical question answering system(CMQA)based on deep learning.For the CMQA system,it is the first step of the system to understand the user ’s Chinese question and convert it into a structured query sentence.It is also a major difficulty for the entire Q & A system.In order to solve this problem,an automatic generation method based on Seq2 Seq and a generation method based on information extraction are proposed;the former mainly solves the question understanding that contains multiple semantic relations,and is composed of three main components,namely a generator,a learner and an interpreter.The role of the generator is to generate the data set required for the experiment,specifically for the use of the query template specifically written in advance for 13 relationships or attributes in the medical knowledge base and the disease entity under the Chinese medical knowledge base disease classification to generate the experiment.Data set;the role of the learner is to learn the conversion rules of Chinese natural language questions to SPARQL query sentences from the training data set;the interpreter is to use the deep learning model learned by the learner to generate natural questions to SPARQL.The latter mainly solves the understanding of a single relation question,which is mainly implemented in two steps.First,the entity recognition model based on Bi LSTM is used to extract the entities in the user question;then the classifier based on SVM is used to determine the category of the user question.The class represents the corresponding relationship in the knowledge base;for the above models,experiments are conducted separately.The experiments show that the proposed model has good performance on data sets.Finally,the above model is integrated to realize the CMQA Chinese medical question answering system,which can understand the problem well and generate SPARQL query,which proves that this research has certain application value.