OJ Intelligent Training System based on Artificial Intelligence for Hierarchical Personalized Teaching

This paper introduces a new feature of the OJ system at Donghua University: intelligent training. Compared to other OJ systems, our OJ has the following features: 1) A clustering algorithm is used to classify the problems. When a student completes a category of problems, he or she passes a level and is awarded the corresponding score. 2) The system intelligently determines whether a student can pass the level or not. Generally, students with higher abilities need to complete fewer problems but receive higher scores. 3) The system uses a classification algorithm to determine whether plagiarism has occurred based on the code submitted by the student and the behavioral characteristics of the student at the time of submission, thus preventing plagiarism as much as possible. 4) The system uses a BP neural network to predict students' final exam scores, and then uses that predicted value as a point for reflecting ability to enhance students' sense of achievement in completing the practices. The OJ intelligent training model, based on the above features, allows students to train themselves at different levels according to their ability. Weaker students can practice more at lower levels to strengthen their foundation, while stronger students can practice more at higher levels to compete in competitions. This achieves the goal of personalized teaching at different levels and achieves better teaching results.