Search papers, labs, and topics across Lattice.
This study investigates cross-rubric generalization in automated essay scoring (AES), focusing on the ability to score essays using previously unseen rubrics that assess different aspects of critical thinking. By employing a fine-tuning framework that incorporates rubric-agnostic intermediate representations (traits) and targeted supervision, the authors demonstrate a 5.0% improvement in macro F1 scores in challenging scenarios where both target rubrics and essays are unseen during training. The findings indicate that leveraging trait-based structures and enhanced supervision can significantly enhance the generalization capabilities of AES systems.
Trait-based representations can boost essay scoring performance by 5% even when faced with entirely new rubrics.
Automated essay scoring (AES) research has largely focused on cross-prompt generalization, where essays from unseen prompts are scored while the scoring criteria are typically held constant. In practice, however, educators may revise or even introduce new rubrics in their scoring task, to evaluate different aspects of essays. We study cross-rubric generalization: training on essays labeled under one set of rubrics and evaluating on previously unseen rubrics, which target different aspects of the essay. We use a Large Language Model (LLM) fine-tuning framework with two components: rubric-agnostic intermediate representations, called traits, and target-essay supervision under seen rubrics during training. On an AES dataset augmented with multiple rubric-defined labels of student critical thinking skills, we find that traits improve macro F1 by 5.0% over a baseline without traits in the hardest setting, where both target rubrics and target essays are unseen during training. We further find that increasing target-essay supervision improves performance, with our best fine-tuned open-source Llama-based model outperforming GPT-5-mini prompting by 2.1% macro F1 and trailing GPT-5 by 1.9%. These results show that trait-based intermediate structure and controlled supervision improve generalization to unseen rubrics.
