Personalising Learning Through Emotions to Improve Educational Experience: Sentellect

Abstract

This document introduces Sentellect, a web-based adaptive learning system powered by AI, aimed at improving educational experiences for 12th-grade Mathematics (PTB) students through the incorporation of emotional state analysis into content delivery. The primary aim is to go beyond conventional performance metrics by evaluating a student's cognitive and learning abilities through an extensive survey assessment. A Random Forest classifier, trained and validated with an accuracy of 96.60%, is employed to forecast student distress levels (Low, Moderate, High) using these inputs. This forecast, combined with performance metrics, powers the personalization engine, creating a customized learning pathway. To foster trust and acceptance, the system includes an Explainable AI (XAI) element to deliver transparency for every learning suggestion. Additionally, a unified Chatbot driven by a Large Language Model (LLM API) provides immediate assistance and emotional backing. The completed Sentellect web application, created with Python Flask and HTML/CSS, showcases a thorough solution that adjusts content complexity, speed, and assistance methods in real-time, promoting a more supportive, interactive, and efficient educational setting.