Intelligent auxiliary evaluation method for higher education quality based on data mining technology

Higher education quality assessment faces challenges such as strong heterogeneity of multi-source data, complex feature dimensions, and the need for stable and reliable evaluation results. The large-scale structured data processing capabilities of data mining technology (SLIQ) enable precise screening of key feature factors, ensuring stable and reliable assessment outcomes. Therefore, this study investigates an intelligent auxiliary evaluation method for higher education quality based on data mining technology. By constructing a multimodal educational knowledge graph, this approach integrates multi-source educational data. A combination of LightKGCN, LightGCN, and an improved RippleNet graph neural network extracts high-order features from this knowledge graph, generating a structured feature table. The SLIQ data mining algorithm then identifies key factors influencing educational quality from this structured feature table. A random forest evaluation model is built using these key factors to achieve intelligent-assisted higher education quality assessment. The results demonstrate that this method can screen out 10 key influencing factors from 28-dimensional features, with student interaction intensity and resource relevance being the most significant (information gains of 0.142 and 0.138, respectively). The classification error rate of the random forest evaluation model based on key factors on the test set was reduced to about 0.013 (standard deviation 0.002). The evaluation results highly correlate with actual educational quality ratings, validating the accuracy, stability, and interpretability of this method.