Arg-LLaDA: Argument Summarization via Large Language Diffusion Models and Sufficiency-Aware Refinement

2025-07-25

AI Summary

To address factual inconsistencies and structural rigidity arising from single-pass generation in complex multi-perspective debate summarization, this paper proposes an iterative summarization framework based on large language diffusion models. The core innovation is a sufficiency-guided remasking mechanism: a mask controller dynamically identifies redundant or insufficient segments, while a sufficiency verification module drives multi-round regeneration and refinement to jointly optimize content faithfulness, structural coherence, and expressive conciseness. The method integrates diffusion-based generation, dynamic masking control, and differentiable sufficiency evaluation, enabling fine-grained content revision. On two benchmark datasets, our approach achieves state-of-the-art performance on 7 out of 10 automated metrics and significantly improves human-evaluated coverage (+12.3%), faithfulness (+15.6%), and conciseness (+10.8%).

Technology Category

• Natural Language Processing: Summarization
• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Argument summarization aims to generate concise, structured representations of complex, multi-perspective debates. While recent work has advanced the identification and clustering of argumentative components, the generation stage remains underexplored. Existing approaches typically rely on single-pass generation, offering limited support for factual correction or structural refinement. To address this gap, we introduce Arg-LLaDA, a novel large language diffusion framework that iteratively improves summaries via sufficiency-guided remasking and regeneration. Our method combines a flexible masking controller with a sufficiency-checking module to identify and revise unsupported, redundant, or incomplete spans, yielding more faithful, concise, and coherent outputs. Empirical results on two benchmark datasets demonstrate that Arg-LLaDA surpasses state-of-the-art baselines in 7 out of 10 automatic evaluation metrics. In addition, human evaluations reveal substantial improvements across core dimensions, coverage, faithfulness, and conciseness, validating the effectiveness of our iterative, sufficiency-aware generation strategy.

Problem

• Generating concise structured summaries of complex debates
• Improving argument summarization via iterative refinement
• Enhancing faithfulness and conciseness in summary outputs

Innovation

• Large language diffusion framework for summaries
• Sufficiency-guided remasking and regeneration
• Flexible masking controller with sufficiency-checking