A Programming Learning Platform with Misconception Diagnosis

Programming novices often encounter difficulties in mastering fundamental concepts due to common misconceptions, which can manifest as syntax errors, logical flaws, and incorrect problem-solving strategies. To address this challenge, this study develops a programming learning platform that provides real-time diagnosis and feedback on misconceptions. Students can submit code to verify its correctness and receive immediate feedback on identified misconceptions, while instructors can monitor learning progress and adjust instructional strategies accordingly. The platform utilizes data mining techniques to extract code features, which are then analyzed using spectral clustering algorithms to identify misconception symptoms and categorize them into specific types. By mapping the relationships among various misconceptions, the platform enhances instructors’ understanding of students’ learning obstacles. Future work will focus on designing more guided feedback based on misconception patterns to better support students in correcting misunderstandings and improving their programming skills.