Tapping the Potential of Coherence and Syntactic Features in Neural Models for Automatic Essay Scoring
This work addresses the challenge of improving scoring accuracy in educational assessments, though it appears incremental as it builds on existing neural and hybrid models.
The paper tackles the problem of automatic essay scoring by proposing a novel approach that extracts coherence features using prompt-learning NSP and augments BERT-based models with syntactic feature dense embeddings, achieving state-of-the-art performance for long essays and hybrid methodologies.
In the prompt-specific holistic score prediction task for Automatic Essay Scoring, the general approaches include pre-trained neural model, coherence model, and hybrid model that incorporate syntactic features with neural model. In this paper, we propose a novel approach to extract and represent essay coherence features with prompt-learning NSP that shows to match the state-of-the-art AES coherence model, and achieves the best performance for long essays. We apply syntactic feature dense embedding to augment BERT-based model and achieve the best performance for hybrid methodology for AES. In addition, we explore various ideas to combine coherence, syntactic information and semantic embeddings, which no previous study has done before. Our combined model also performs better than the SOTA available for combined model, even though it does not outperform our syntactic enhanced neural model. We further offer analyses that can be useful for future study.