Decision-Level Ordinal Modeling for Multimodal Essay Scoring with Large Language Models

TL;DR

This paper introduces Decision-Level Ordinal Modeling (DLOM), a novel approach for multimodal and text-only automated essay scoring that explicitly models ordinal trait scores using large language models, improving accuracy and interpretability.

Contribution

The paper proposes DLOM, a new method that makes essay scoring an explicit ordinal decision, with specialized modules for multimodal and text-only scenarios, enhancing performance over existing generation-based methods.

Findings

DLOM outperforms generation-based baselines on the EssayJudge dataset.

DLOM-GF improves scores when modality relevance varies.

DLOM-DA enhances performance on text-only benchmarks.

Abstract

Automated essay scoring (AES) predicts multiple rubric-defined trait scores for each essay, where each trait follows an ordered discrete rating scale. Most LLM-based AES methods cast scoring as autoregressive token generation and obtain the final score via decoding and parsing, making the decision implicit. This formulation is particularly sensitive in multimodal AES, where the usefulness of visual inputs varies across essays and traits. To address these limitations, we propose Decision-Level Ordinal Modeling (DLOM), which makes scoring an explicit ordinal decision by reusing the language model head to extract score-wise logits on predefined score tokens, enabling direct optimization and analysis in the score space. For multimodal AES, DLOM-GF introduces a gated fusion module that adaptively combines textual and multimodal score logits. For text-only AES, DLOM-DA adds a distance-aware regularization term to better reflect ordinal distances. Experiments on the multimodal EssayJudge dataset show that DLOM improves over a generation-based SFT baseline across scoring traits, and DLOM-GF yields further gains when modality relevance is heterogeneous. On the text-only ASAP/ASAP++ benchmarks, DLOM remains effective without visual inputs, and DLOM-DA further improves performance and outperforms strong representative baselines.