Research On Medical Automatic Question Answering System Based On Attention Mechanism And Convolutional Depth Structure Semantic Model

With the rapid development of the Internet,information data has increased ex-ponentially.How to effectively use this information has become a hot spot of people’s attention,and then automatic question and answer technology has come into people’s eyes.Automatic question answering systems are divided into open domains and re-stricted domains according to the fields involved in the content.The automatic question answering system based on the open domain has received extensive attention from all walks of life,and has made certain research progress,but it cannot be directly applied to the restricted domain.The reason is that the knowledge and professionalism of the restricted domain is strong,and the automatic question answering system based on the open domain cannot correctly judge and identify it.Therefore,it is also indispensable for the research of restricted domain automatic question answering system.The key link of the domain-based automatic question answering technology is text matching and answer extraction.The essence is to sort the candidate answers by calculating the similarity between the question and answer pairs,and return the answer with the highest similarity to the user.In order to make full use of the semantic information of the text and improve the accuracy of the model,this paper has done the following work:First,Build a database.The question and answer pairs are crawled from the pro-fessional medical platform,after data preprocessing,redundant characters and non-compliant queries and answers are removed,the medical field database is constructed,and the problem of insufficient professional knowledge is effectively solved.Second,ap-ply the convolutional depth structure semantic model to the medical field.Since the bag of words model used in the deep structure semantic model cannot take into account grammatical information,combining the convolutional neural network with the model can make up for the lack of the deep structure semantic model to consider the position information and word order information of the words,so that the model can extract latent semantics information.Third,a deep learning method based on attention mecha-nism and convolutional deep structure semantic model is proposed.On the basis of the convolutional depth structure semantic model,the attention mechanism is introduced to make the model more sensitive to text information.It can ignore the irrelevant information of the text and focus on the effective information,expand the weight of the effective information,and improve the accuracy of the model.At last,experimental verification.Experiment with the model and compare and analyze the experimental data.Simultaneously,the model is simulated and tested to verify the validity of the model.This paper conducts experiments on the basis of self-built data sets and finds that the attention mechanism and convolutional deep structure semantic model are compared with the deep structure semantic model,and indicators such as Acc and MRR have been improved.The returned answers obtained by simulating questions on the model are also reasonable,which verifies the effectiveness of the method in many aspects.