arXiv cs.CL· Xuanxin Wu, Yuki Arase, Masaaki Nagata·· 13 小时前AI 评分31
用 LLM-Judge 自动构建策略对齐训练数据，实现可控文本简化
Edit Policy-Aligned Text Simplification
AI 导读
研究者提出用 LLM-Judge 自动构建策略对齐训练数据的方法，无需昂贵人工标注或平行语料，即可训练适配不同简化策略的模型，覆盖句子级与文档级。小规模开源 LLM 在自动指标和编辑分析中达到或超过 GPT-5，且跨模型家族与规模均有一致提升。该工作已被 AACL-IJCNLP 2026 接收。
正文 · 原文
Abstract:Text simplification aims to modify documents and sentences to make them easier to read and understand while preserving their meanings. Different applications require distinct simplification policies, e.g., lexical-oriented simplification and overall rewriting while trading off detail for simplicity. However, achieving such policy-driven control remains an open challenge. In this work, we introduce a simple yet powerful approach that leverages Large Language Model-as-a-Judge (LLM-Judge) to automatically construct policy-aligned training data, removing the need for costly human annotation or parallel corpora. Our method enables the construction of simplification models that adapt to diverse simplification policies at both the sentence and document levels. Remarkably, small-scale open-source LLMs match or surpass GPT-5, as verified by both automatic metrics and edit analysis. The consistent improvements across model families and sizes demonstrate the robustness of our approach.
| Comments: | Accepted by AACL-IJCNLP 2026 |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2512.06228 [cs.CL] |
| (or arXiv:2512.06228v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2512.06228 arXiv-issued DOI via DataCite |
Submission history
From: Xuanxin Wu [view email]
[v1]
Sat, 6 Dec 2025 00:29:49 UTC (1,947 KB)
[v2]
Wed, 21 Oct 2026 07:53:46 UTC (2,224 KB)
来源：arXiv cs.CL · arxiv.org