Research And Application Of Key Technology Of Intelligent Teaching Assistant Robot

In response to the national call to cultivate applied talents,more and more colleges and universities pay more attention to the cultivation of students’ practical ability,which is embodied in the practice link which is deeply combined with the theory course,through which students’ practical ability can be continuously improved.Course assistant is a common teaching mode in Colleges and universities,which plays a key role in the mastery of students’ theoretical knowledge and the improvement of their practical ability,and has become an indispensable part in the daily education of colleges and universities.However,there are some key problems to be solved in traditional human assistant teaching.For example,low efficiency,time and space limitations,limited level and so on.In order to improve the efficiency of classroom practice in Colleges and universities.In this thesis,an intelligent teaching assistant is developed based on neural network and micro service technology.The main contents are as follows:(1)The whole framework of intelligent teaching assistant is designed by using micro service technology.In order to reduce the interference between modules,in this thesis,the intelligent teaching assistant is designed as three modules and two centers.The three modules are practice guidance module,intelligent question answering module and speech recognition module.The two centers are configuration center and registration center.The design of extensibility lays a good foundation for service extension.(2)Completed the design and implementation of three modules.In the practice guidance module,this thesis uses websocket and tornado technology to complete the command verification,configuration loading,command execution and other sub modules.In the intelligent question answering module,this thesis introduces how to use crawler technology to crawl the data set in the limited domain,and based on it,completes the enhanced training of SIMNET semantic matching model and the construction of corpus with inverted index.In addition,this thesis adds the Q&a log storage and analysis sub module in the module,and gives a log analysis method combined with map reduce programming model and K-means clustering algorithm.In the speech recognition module,this thesis uses Python to transform the audio signal into spectrogram,and as the input of dfcnn acoustic model,improves the recognition accuracy of the model,and proposes the combination of dfcnn acoustic model and ngram language model to complete the construction of speech recognition system.(3)Completed the implementation and testing of intelligent teaching assistant for big data field.In this thesis,in the design and implementation After the implementation of intelligent teaching assistant,it has been successfully applied to the course "massive information processing technology and practice" of Beijing University of Posts and telecommunications,and completed the function test and performance test.In addition,this thesis compares with the traditional question answering system and speech recognition system.The experiment shows that the intelligent teaching assistant based on SIMNET and dfcnn has a good effect.Finally,the analysis of the question and answer log also points out the weakness of students’ knowledge theory,which has a good role in promoting the teaching effect.