An Intelligent Medical Question Answering System Using Natural Language Processing

In recent years,with the rapid development of Internet,the content and information on the Internet are exploding,and it will be more difficult for non-professionals to obtain effective medical information from search engines.At the same time,because the level of medical construction in my country cannot keep up with the needs of patients for medical treatment,the contradiction between doctors and patients has not been fundamentally resolved,so the problem of difficult consultations is widespread.With the advent of the era of big data and artificial intelligence,people have become more agile and efficient in acquiring all kinds of information.It is an urgent and important issue to accelerate the realization of informatization in the medical field.In order to make it convenient for patients to consult a doctor when they need to see a doc-tor,and to obtain professional and humanized answers,this paper proposes a retrieval question answering system in the medical field.This paper using Natural Language Processing tech-nologies and mainly focuses on building an IR-based question answering system that combines a pretrained language model and entity matching.The main work of this paper includes:1.This paper proposes a text semantic matching model FusionLM that combines a pre-trained language model and a siamese network.This model introduces the siamese network structure on the basis of the BERT model,and integrates multiple semantic similar features in the question answer pairs,and learns deeper matching information from question answer pairs.This paper conducts experiments with FusionLM on public datasets,and the effect of the model has reached the advanced level in the answer selection task,which can better assist in completing the answer selection module in the question answering system.2.This paper proposes an enhanced algorithm for answer selection based on entity match-ing.This paper introduces the entity matching module into the traditional IR-based question answering system architecture,and adds named entity recognition and knowledge graph to the question answering system as a scoring basis to assist in selecting the best answer.In this paper,through experiments on public datasets,the text matching model with entity matching module can effectively improve the accuracy of answer selection in the task of answer selection.3.This paper builds an intelligent medical question answering system and develops a cor-responding online platform.This paper collects a large number of effective data and puts the above two algorithms into practice.Based on the retrieval question answering system archi-tecture,a scoring mechanism that combines multiple information is designed,and a complete medical question answering system is constructed.This paper has verified the validity and ac-curacy of the answers through manual evaluation of the question answering system.Based on the question answering system architecture,this paper develops a real-time online intelligent medical question answering system platform for patients and doctors to inquire about medical problems.