Research On Automatic Reply Method Of Course Interactive Communication Question

Automatic reply means that the system can automatically provide an answer to the user's question.In automatic reply tasks,the coincidence rate and similarity of vocabulary between most questions and answers is not high.It is difficult to solve using only word matching and feature extraction methods.In recent years,deep learning has been widely used in natural language processing and has achieved good results.The automatic reply method for interactive communication problems studied in this paper has a wide range of applications and is currently a research hotspot.The existing question answering system includes a community question answering system,a knowledge answering system and a chat robot.These systems are based on the analysis of the questions raised by the users in the case of a certain knowledge reserve,to obtain candidate answers,and then select the best answer from the candidate answers to return to the user.This topic mainly studies the method of automatic reply to the problem of interactive communication in the course.Its core issues include two parts,The first part is based on the question asked by the questioner to get a similar problem set,the second part is to sort all the candidate answers to get the best answer to the user.This paper uses word vector technology to calculate the similarity of questions and using the deep learning model to build a candidate answer selection model,Changed traditional methods for similarity calculation and answer extraction based on simple word matching and feature extraction,This article is mainly from the following two aspects:(1)Research on the similarity calculation method of interaction problem based on word vector,By introducing the current popular word vector model into problem similarity calculation,the semantic representation of words in sentences is enhanced.A similarity degree calculation method was designed to calculate the similarity between questions by combining cosine similarity.Finally,comparative experiments show that the similarity calculation method of the sentence sentence designed in this paper is superior to the traditional similarity calculation method in accuracy and recall rate.(2)Build an answer selection model based on deep learning,This paper builds the answer selection model based on the bidirectional LSTM model.And the Attention mechanism was introduced to improve the model's effectiveness.Based on the characteristicsof the curriculum interactive communication platform in this study,non-text features are incorporated into the answer selection model.Finally,an answer selection model suitable for the course communication platform was constructed to extract the best answer from the candidate answers.(3)System design and implementation of the course exchange platform,From the practical application point of view,Through requirements analysis,system design,system implementation and other processes,and the problem similarity calculation method and the answer selection method designed in this paper are used to realize the automatic reply of the course interactive communication problem.