Personalized AI-driven Teacher’s Comment Model for Nigerian Secondary School Students’ Result

The paper presents a personalized AI-driven system for generating teacher comments on Nigerian secondary school report cards. It addresses the challenge of producing individualized, high-quality feedback in overcrowded classrooms. Using 285 synthetically generated student records—including subject scores, cognitive and psychomotor ratings—the authors fine-tuned a FLAN-T5 transformer model to generate comments from structured data. The methodology involved prompt engineering, data cleaning, and model optimization via grid search and cross-validation. Although dummy labels were used, the training validated the technical pipeline and model compatibility. Results showed functional learning behavior, with future work aimed at expanding the dataset, incorporating multilingual support, and refining cultural alignment. The study demonstrates that lightweight, interpretable AI models can reduce teacher workload while maintaining feedback quality in resource-constrained educational settings.