Research And Implementation Of Answer Selection Method For Medical Filed

Due to the increase in the pressure of offline medical consultations and the increase in the number of online consultations,there are more and more researches on medical automatic question answering systems.Such systems bring convenience to patients and also serve as auxiliary tools to alleviate pressure of doctors.The rapid development of the Internet and deep learning technology has also provided a solid foundation for the development of the automatic question answering system,and the answer selection technology is one of the most important technologies to realize the question answering system.In order to alleviate the medical pressure and improve the precision of matching answers with user questions,the research on answer selection methods have been carried out in the medical field.Compared with the question-answer matching task in the English open field,the task in the Chinese professional medical field is more challenging.In view of the complexity and diversity of Chinese semantics and medical data,most researchers focus on designing complex neural networks to explore the deeper text semantics,and this kind of idea is relatively simple.At the same time,the more complex neural network model is more vulnerable to the impact of small disturbances,resulting in poor robustness ability of the model.To this end,a question-answer matching model(AT-RoBERTa)based on RoBERTa and adversarial training is proposed.A bidirectional pre-training encoder is used in this model to capture the semantic information of question and answer to obtain the corresponding vector representation,and then the disturbance factor is added to the word embedding representation to generate the adversarial sample,and then the initial sample and the adversarial sample are input into the model for adversarial training,and finally the classification prediction is completed through the linear layer.According to the professional process of software engineering for system development,a medical question answering system based on the proposed model is designed and implemented.The experimental results of AT-RoBERTa on medical datasets show that the introduction of adversarial training can effectively improve the performance of the question-answer matching model,which provides a new idea for the realization of the automatic question-answer function.At the same time,the operational effectiveness of the designed and developed question answering system also proves the feasibility of the model in practical application,which can assist doctors in their work,so as to achieve the goal of reducing the pressure of online consultation and offline consultation.