Abstract—Automated Essay Scoring (AES) systems have achieved notable success in evaluating semantic content using deep learning models. However, they often fail to provide explicit, interpretable feedback on essay structure. This paper presents a hybrid neural-symbolic approach that integrates Bidirectional Encoder Representations from Transformers (BERT) for semantic analysis with a Hidden Markov Model (HMM) for explicit structural modeling. Unlike ensemble methods, our system employs a tightly coupled, weighted integration scheme (70% content, 30% structure) optimized through validation experiments. The HMM component models essay organization as a sequence of rhetorical states Introduction, Body, and Conclusion offering transparent feedback on logical flow. We explicitly position this work as a methodological proof-of-concept validation study.
 
Keywords—Automated Essay Scoring (AES), Bidirectional Encoder Representations from Transformers (BERT), Hidden Markov Model (HMM), interpretable AI, educational technology, neural-symbolic systems


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).