Study On Intelligent Teaching System Featuring Breakdown Of Knowledge Points

In recent years,especially after the outbreak of novel coronavirus,network teaching has attracted wide attention from society and schools,and a large number of network teaching software and websites have emerged.Indeed,the development of the computer,the Internet,such as artificial intelligence technology to make education entered the information era,the teaching mode and teaching methods have changed,even if not in school students can also hear the teacher through the network,anytime and anywhere browse downloading of teaching resources,but the informatization education means,after all,is a kind of auxiliary tools,most of the time don’t brings to the teaching quality and sometimes even weaken the humanization and flexibility characteristics of daily face-to-face classroom teaching,such as too fast rhythm makes the teacher and the students often disconnected;Too many network teaching resources make it difficult for students to choose,but waste valuable learning time and so on.The laboratory plans to use streaming media technology and artificial intelligence technology to better integrate network teaching means into daily classroom teaching,so that they can give full play to their respective advantages,learn from each other,and build an intelligent assisted teaching system.This system is purchased by the teaching website running on the server and the APP running on the personal terminal of teachers and students.It can record the daily class process and supervise the learning status of students at a low cost.It can also promote teacher-student interaction through live-streaming barrages and realize personalized learning with the help of intelligent recommendation.This paper undertakes a large number of research and development work,the main work and achievements are as follows.This paper develops a teaching assistant website based on WebRTC streaming media technology,RTMP real-time message transmission protocol and FFMPEG multimedia framework.The website has two core functions,one is to assist the class function.Through the website,teachers can record each class in the way of screen recording in the background,and transmit the recorded video to the student terminal in real time through the live broadcast technology.Finally,the recorded complete class video will be saved on the cloud server.In addition,the use of chat room technology to achieve group communication functions,including not only sending text,pictures and documents and other basic functions,but also including attendance statistics,homework scoring and other special functions,in order to reduce the burden of teachers,improve teaching efficiency.Second,the big data function.You can collect,organize,search and recommend daily classroom teaching videos and exercises from different schools and teachers day after day.In order to manage the teaching video efficiently,this paper proposes a method that can quickly edit and organize the video according to the knowledge points.The method first uses the speech recognition technology to convert the speech in the video into text,and generates subtitle files.Then it divides the whole teaching video into sections or extracts the knowledge points with the help of the time period when the keywords appear in the subtitle files,and produces short teaching videos with subtitles.Using this kind of short video which is divided into sections according to knowledge points,students can search and learn relevant knowledge points according to their interest points,rather than the whole video,thus greatly reducing the burden of students and improving learning efficiencyIn order to efficiently manage the problem bank,this paper proposes a processing method that can quickly classify the problem bank according to knowledge points.This method uses TF-IDF algorithm to extract features from the exercise text and construct vectors,and uses Naive Bayes and support vector machine classification algorithm to classify various types of questions in the data set.In this paper,a set of exercise data set collected manually is trained by supervised learning method.The experiment shows that the classification performance of Naive Bayes is better than that of SVM in the case of small exercise data set.If enough exercises are collected,the classification effect will be more excellent.In order to realize the personalized learning,the paper fully analyzes the teaching website users and recommended object characteristics,and targeted to build its dynamic and static characteristics describe vector,implements the four kinds of recommendation algorithm,respectively is recommended according to the personal interest,according to individual test level recommended hot,according to friends hobby is recommended,and according to the website content is recommended.Hacker News algorithm is recommended according to the hot content.This algorithm takes into account both the popularity score of teaching resources and the number of days they are uploaded,and recommends the hottest and latest resources to students and teachers in a timely manner.And test level recommended comprehensive consideration according to the individual subject degree of difficulty and students’ mastery of knowledge points,this paper proposes a combination of initial difficulty setting and dynamic adjustment calculation method of the difficulty of problem sets,and build the tree of knowledge,to make the teaching video website and problem sets can be linked through a web of knowledge,to better make personalized recommendation for students or teachers.Experimental tests show that the teaching website developed in this paper has cross-platform characteristics and can run on Windows laptop,tablet computer and Android mobile phone.The teaching website developed in this paper works stably,and the functions such as screen recording and group communication all meet the preset requirements.With the continuous operation of the website day after day,more and more teaching resources will be accumulated in the website,and the recommendation process of the recommendation module will be more accurate and more personalized.