Enhancing Student Feedback and Assessment with Graph Attention Networks in Higher Education

The current approaches to student feedback and evaluation in higher education frequently fail to effectively capture and evaluate the varied and intricate feedback data. This study aims to tackle this problem by introducing an innovative approach that utilises Graph Attention Networks (GATs) to improve the analysis and assessment of student comments. The suggested system adopts a methodical approach, commencing with the gathering of data from extensive educational datasets obtained from Kaggle. By carefully analysing the data and selecting relevant features, the system creates graph representations that accurately depict the complex connections between students, courses, and feedback. The GAT model is subsequently applied and trained on these graphs, utilising a dynamic updating technique to consistently enhance performance. A comparative investigation reveals that the GAT-based system exhibits superior performance compared to standard models like Random Forest (RF) and Support Vector Machine (SVM). The findings demonstrate a significant enhancement in accuracy (92.7%), precision (91.3%), recall (93.2%), and F1score (92.2%) when compared to RF and SVM. This sophisticated analytical framework improves the precision of student performance assessments and also offers more profound insights into educational outcomes, presenting a more efficient and adaptable option for higher education institutions. This study establishes a new standard in the sector by showcasing exceptional performance indicators and a thorough methodology for managing student feedback data.