MAPLE: A Meta-learning Framework for Cross-Prompt Essay Scoring

TL;DR

MAPLE is a meta-learning framework that improves cross-prompt automated essay scoring by learning transferable representations, achieving state-of-the-art results across multiple datasets and languages.

Contribution

This paper introduces MAPLE, a novel meta-learning approach using prototypical networks to enhance generalization in cross-prompt essay scoring.

Findings

MAPLE outperforms strong baselines by 8.5 and 3 points in QWK on ELLIPSE and LAILA datasets.

MAPLE achieves state-of-the-art performance on ELLIPSE and LAILA datasets.

MAPLE improves scoring performance on heterogeneous prompt datasets like ASAP.

Abstract

Automated Essay Scoring (AES) faces significant challenges in cross-prompt settings, where models must generalize to unseen writing prompts. To address this limitation, we propose MAPLE, a meta-learning framework that leverages prototypical networks to learn transferable representations across different writing prompts. Across three diverse datasets (ELLIPSE and ASAP (English), and LAILA (Arabic)), MAPLE achieves state-of-the-art performance on ELLIPSE and LAILA, outperforming strong baselines by 8.5 and 3 points in QWK, respectively. On ASAP, where prompts exhibit heterogeneous score ranges, MAPLE yields improvements on several traits, highlighting the strengths of our approach in unified scoring settings. Overall, our results demonstrate the potential of meta-learning for building robust cross-prompt AES systems.