Go to ACL ARR 2025 May homepageCAFES: A Collaborative Multi-Agent Framework for Multi-Granular Multimodal Essay Scoring
Abstract: Automated Essay Scoring (AES) is crucial for modern education, particularly with the increasing prevalence of multimodal assessments.
However, traditional AES methods struggle with evaluation generalizability and multimodal perception, while even recent Multimodal Large Language Model (MLLM)-based approaches can produce hallucinated justifications and scores misaligned with human judgment. To address the limitations, we introduce CAFES, the **first collaborative multi-agent framework specifically designed for AES**.
It orchestrates three specialized agents: an initial scorer for rapid, trait-specific evaluations; a feedback pool manager to aggregate detailed, evidence-based strengths; and a reflective scorer that iteratively refines scores based on this feedback to enhance human alignment. Extensive experiments, using state-of-the-art MLLMs, achieve an average relative improvement of 21% in Quadratic Weighted Kappa (QWK) against ground truth, especially for grammatical and lexical diversity. CAFES paves the way for an intelligent multimodal AES system.
Paper Type: Long
Research Area: Multimodality and Language Grounding to Vision, Robotics and Beyond
Research Area Keywords: Automated Essay Scoring, Multi-Agent System, Multimodal Large Language Models
Contribution Types: Model analysis & interpretability
Languages Studied: English
Keywords: Automated Essay Scoring, Multi-Agent System, Multimodal Large Language Models
Submission Number: 294
Loading