Automated Models in Educational Assessment: A Comprehensive Survey

This study presents a review of automated grading systems that address scalability, efficiency, and consistency challenges in educational assessments. The framework utilizes cutting-edge natural language processing models, including Paragraph Vector, BERT, and a hybrid LDA-BERT approach, to evaluate student responses with high accuracy. By integrating semantic similarity and topic alignment, the system outperforms traditional grading methods, achieving improved contextual understanding and relevance. Designed for large-scale assessments, the system provides timely feedback, reduces educator workload, and ensures scalability. Potential future applications include multilingual assessment integration, real-time feedback, and adaptive learning environments, which will enhance its versatility across various educational contexts. This work paves the way for the development of fair, scalable, and AI-driven educational tools that contribute to more effective and equitable learning outcomes.