Enhancing Attainment and Quality Assurance in Higher Education with Machine Learning-Driven Smart Pedagogy

Conventional educational methods struggle to deliver personalised learning experiences, resulting in limitations in student engagement and achievement. This study introduces a hybrid model combining Knowledge Graph Neural Network and Natural Language Processing (KGNN-NLP) to improve higher education teaching. The system generates a knowledge network from educational data, utilises natural language processing for contextual comprehension, and enhances learning paths through reinforcement learning. A comparative examination with conventional models reveals greater performance, with 94.8% accuracy, a 0.92 F1-score, and a 0.08 MAE. The system surpasses baseline models like Decision Trees, SVMs, and RNNs, guaranteeing more flexibility, increased retention, and an 18% reduction in dropout rates. The suggested method connects structured and unstructured learning data, allowing automated curriculum development, customised lesson suggestions, and predictive analytics for student achievement. This approach, compared to current models, integrates deep learning with knowledge reasoning, therefore establishing a scalable and intelligent learning framework. Future enhancements will investigate multi-modal learning, real-time adaptive evaluations, and interdisciplinary applications.