Intelligent teaching quality assessment in higher education using IoT and graph neural networks

Ensuring high teaching quality in higher education institutions is essential for improving student learning outcomes and institutional performance. Traditional evaluation methods rely heavily on surveys and manual assessments, which often lack real-time insights, objectivity, and scalability. This research aims to develop an intelligent teaching quality evaluation using Internet of Things (IoT) technologies to enhance student performance, engagement, and interactions. The model utilizes teaching quality classroom data with 4000 rows, including attendance records, environmental parameters, learning engagement metrics, assessment scores, and interaction logs from higher education institutions. Pre-processing techniques include data cleaning, min-max normalization, and missing value handling. Principal Component Analysis (PCA) decreases dimensionality and extracts useful characteristics from IoT data. This research proposes an Adaptive Teaching–Learning-Based Optimization Guided Attention Graph Neural Network (ATLBO-AGNN) model designed to predict and evaluate teaching quality in higher education. The ATLBO for hyperparameter optimization of teaching strategies to maximize learning outcomes, while AGNN is used to learn complex relationship of students, instructors, and learning activities. Experiments using Python have shown that the suggested ATLBO-AGNN model outperforms the other model and attains higher accuracy results, with a range of 96% to 98%. The proposed ATLBO-AGNN method effectively enhances teaching quality assessment by leveraging IoT data and relationship-aware knowledge systems, supporting continuous improvement in higher education systems.