Research And Implementation Of Scratch Online Programinglearning System

Scratch is a children's programming learning tool from MIT.Since Scratch is no longer necessary to remember the complicated grammar and variable,it only needs to be programmed in the form of dragging blocks,Scratch is becoming the most popular way for young people to develop logical thinking skills and basic programming literacy.However,Scratch only provides a programming environment.From the perspective of teaching,there is no effective means to track the learning process and understand the learning behavior.And it is difficult to achieve the goal of teaching guidance and optimizing teaching effects.This paper introduce big data analysis technology into the online programming learning system to achieve the automatic multi-label classification for Scratch projects.And it also achieve a goal that automatic tracking and modeling analysis of the course learning process.It will help to provide personalized learning services and improve teaching effects.Based on the analysis of the mainstream Scratch online programming system function and learning behavior analysis technology,this paper first designs the Scratch online programming learning system architecture that can automatically track the course learning process.Furthermore,according to the automatic classification problem of Scratch projects,a multi-label classification model named LSRF based on tag correlation is proposed.Then we use the dataset which is crawled on Scratch official website to train the classification model of Scratch projects.Aiming at the problem of online learning behavior analysis,this paper proposes a method of serializing learning behavior data and a similar behavior clustering algorithm based on graph model.Then,this paper realizes learning state analysis based on user behavior clustering.Subsequently,the implementation and test plan of each module of the system are introduced in detail.The experimental analysis shows that the Scratch multi-label classification model proposed in this paper has improved the classification accuracy and algorithm running time when compared with other models.In the modeling of online learning behavior,the students'learning behaviors were successfully clustered.Then,we analyzes the differences in the learning effects of different classes of students in terms of behavior.The application shows that the classification algorithm and behavior model proposed in this paper have good practicability.