AI-generated academic assesment portal with performance tracking

Abstract

Artificial intelligence revolutionized numerous digital education operations yet the assessment of academic performance continues to prove especially difficult to overcome. Unfortunately, static question banks and rigid answer matching systems, currently used in making the assessment, can’t provide personalized learning experiences. These systems have difficulty acknowledging proper semantic responses and frequently misidentify them. This paper describes an academic assessment portal developed by AI technology which combines performance tracking features to solve existing evaluation problems. The system uses the multilingual mT5 model to automate the production of questions that match different domains and contextual requirements. The Bangla Transformer system dedicated to evaluation answers detects properly paraphrased responses that improve testing precision. Student performance directs the platform to automatically adjust questions until each student experiences a suitable learning challenge for their current level. The AI system evaluates student responses by analyzing context which enables it to improve both accuracy and fairness of the assessment process. Students obtain performance-related data about their areas of expertise through performance tracking while automated question generation frees educators to teach without additional paperwork. The platform delivers both robustness and user-friendly interface through the use of Flask, React.js. Initial test results indicate that self-assessment performance tracking platforms outperform other methods, showing an accuracy improvement of 40%-50% due to personalized tracking, adaptive learning, and data-driven feedback mechanisms.