Research And Implementation Of Dialogue System For Communication Courses

In recent years,deep learning has developed rapidly,natural language processing as an important field of deep learning,has been developing continuously.At present,the intelligent customer service used in e-commerce,the answers to user input questions by search engines and the machine translation technology used by translation software are all applications of natural language processing in daily life.Natural language processing is widely used,but it has not been studied deeply in the field of education,and the teaching and auxiliary platforms that have emerged focus on the integration of teaching resources or statistics on student attendance,etc.,there are few applications in auxiliary classroom teaching.This paper focuses on the construction of a dialogue system for communication courses.According to the professional characteristics of communication courses,the dialogue system is constructed by end-to-end method using Bi GRU and attention mechanism.,breaking the limitations of the traditional end-to-end model for long texts.On this basis,a dialogue system combining intent recognition and contrastive learning is proposed to solve the problem of insufficient corpus and make the dialogue system more efficient and accurate.The dialogue system designed in this paper solves students’ confusion about course knowledge in the form of dialogue,assists students to better understand the course content after class,saves teachers’ time in answering questions in class,and improves course efficiency.The main research contents of this paper are as follows:(1)Collecting the test questions and answers on communication principles,construct a communication dialogue corpus,preprocessing such as data cleaning,word segmentation and part-of-speech labeling,and training the word vector according to the dialogue corpus to prepare for subsequent model training.(2)Designing an end-to-end dialogue system,test it with a large open source dataset,evaluate its related performance,and then use the communication corpus experiment constructed in this paper to analyze the effect of the dialogue system.(3)On the basis of the end-to-end model,the intention recognition module is added and the data enhancement of the corpus is enhanced using contrastive learning,finally,the dialogue system constructed by different methods is compared,and the better one is selected.(4)Designing the display interface of the web side,the model deployed in the web side,displaying the conversation content in the form of a chat box,and users can get an immediate reply when they enter questions.