CodeSense AI: An Explainable Behavioral Analytics Framework for Virtual Coding Laboratories Using AI-Based Focus Monitoring and Skill Assessment

Abstract

Practical programming laboratories are essential for developing coding skills, but existing virtual coding platforms primarily evaluate the final code submission and provide limited insight into student engagement during lab sessions. This paper presents CodeSense AI, an explainable behavioral analytics framework for virtual coding laboratories that integrates AI-based focus monitoring with programming skill assessment. The proposed system provides a secure coding environment while continuously tracking behavioral metrics such as typing activity, idle time, tab switching, copy-paste attempts, and webcam-based presence detection. These metrics are analyzed to generate Focus Score and Integrity Score, enabling fair and transparent evaluation of student participation. Additionally, an AI-assisted code review module offers personalized feedback on code quality and programming logic. A teacher dashboard presents real-time activity analytics, engagement reports, and performance summaries to support effective monitoring and assessment. The proposed framework improves the authenticity of programming lab evaluations, enhances student accountability, and provides data-driven insights for educators in modern virtual learning environments.