Research And Implementation Of Automatic Correction Of Programming Work Based On Git

With the development of the Internet,traditional education methods have been impacted,and online learning has become a hot spot for domestic and foreign universities.The traditional way of correcting programming assignments has brought great pressure to teachers and students.One is the large number of students’ homework,the other is the particularity of programming homework,and the third is that students cannot get timely homework results.Now more and more colleges and universities at home and abroad adopt Git to manage homework.Teachers publish homework on Git and set up automated scripts.When students submit code,they can automatically run tests.After reading a large number of documents and researching the code hosting platform OneDev,this article found that there are some common phenomena in the current Git management automatic test electronic programming homework,and it is impossible to conduct staged testing based on the Commit content submitted by the students,so if the students do not complete the tasks all You will not get a sign of success status,which may undermine the confidence of students.This article uses this as a research point to classify electronic programming assignments submitted by students.Taking into account the diversity of Commit content,this article uses multi-label text classification,and builds an automated script based on the classification results to complete automatic corrections.This paper uses the TF-IDF text representation method to represent the code language,and then combines the multinomial naive Bayes classifier to classify the TF-IDF feature values.After experimental analysis,the Commit classifier proposed in this paper has a classification accuracy of95.5%,which is very good.Good classification effect.