A Multi-Agent Framework for Feature-Constrained Difficulty Control in Reading Comprehension Item Generation

2026-05-18

AI Summary

This work addresses the challenge that existing question generation methods for reading comprehension often fail to reliably satisfy predefined difficulty constraints, resulting in questions that deviate from the target difficulty level. To overcome this limitation, the authors propose MAFIG, a novel multi-agent framework that introduces collaborative interaction among large language model agents and a feature evaluator. Through iterative co-optimization guided by a difficulty-calibrated constraint sequence, MAFIG enables monotonic and controllable difficulty adjustment during question generation. Experimental results demonstrate that the proposed approach significantly improves adherence to target difficulty characteristics and outperforms current baselines in both controllability and stability.

Abstract

Recent studies in difficulty-controlled reading comprehension item generation have leveraged large language models (LLMs) to produce items by adjusting difficulty-related features. However, existing methods typically rely on a single-agent prompting approach, which often fails to consistently satisfy specified feature constraints, resulting in items that deviate from the target difficulty level. To address this limitation, we introduce MAFIG, a Multi-agent Framework for Feature-constrained Item Generation, where multiple LLM agents and feature-specific evaluators collaborate to generate and iteratively revise items based on intended constraints. Furthermore, to verify the efficacy of MAFIG in difficulty control, we propose a method for constructing a sequence of feature constraint sets that yield items with monotonically increasing difficulty. Experimental results demonstrate that MAFIG generates items that adhere to target constraints at a significantly higher rate than baselines, achieving robust difficulty control through the difficulty-calibrated constraint sequence.

Problem

• difficulty control
• reading comprehension item generation
• feature constraints
• large language models
• multi-agent framework

Innovation

• multi-agent framework
• feature-constrained generation
• difficulty control
• reading comprehension item generation
• large language models