CARO: Chain-of-Analogy Reasoning Optimization for Robust Content Moderation

Bingzhe Wu, Haotian Lu, Yuchen Mou


Abstract
Current large language models (LLMs), even those explicitly trained for reasoning, often struggle with ambiguous content moderation cases due to misleading "decision shortcuts" embedded in context. Inspired by cognitive psychology insights into expert moderation, we introduce CᴀʀO (Chain-of-Analogy Reasoning Optimization), a novel two-stage training framework to induce robust analogical reasoning in LLMs. First, CᴀʀO bootstraps analogical reasoning chains via retrieval-augmented generation (RAG) on moderation data and performs supervised fine-tuning (SFT). Second, we propose a customized direct preference optimization (DPO) approach to reinforce analogical reasoning behaviors explicitly. Unlike static retrieval methods, CᴀʀO dynamically generates tailored analogical references during inference, effectively mitigating harmful decision shortcuts. Extensive experiments demonstrate that CᴀʀO substantially outperforms state-of-the-art reasoning models (DeepSeek R1, QwQ), specialized moderation models (LLaMA Guard), and advanced fine-tuning and retrieval-augmented methods, achieving an average F1 score improvement of 24.9% on challenging ambiguous moderation benchmarks.
Anthology ID:
2026.findings-acl.1442
Volume:
Findings of the Association for Computational Linguistics: ACL 2026
Month:
July
Year:
2026
Address:
San Diego, California, United States
Editors:
Maria Liakata, Viviane P. Moreira, Jiajun Zhang, David Jurgens
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
28880–28892
Language:
URL:
https://aclanthology.org/2026.findings-acl.1442/
DOI:
Bibkey:
Cite (ACL):
Bingzhe Wu, Haotian Lu, and Yuchen Mou. 2026. CARO: Chain-of-Analogy Reasoning Optimization for Robust Content Moderation. In Findings of the Association for Computational Linguistics: ACL 2026, pages 28880–28892, San Diego, California, United States. Association for Computational Linguistics.
Cite (Informal):
CARO: Chain-of-Analogy Reasoning Optimization for Robust Content Moderation (Wu et al., Findings 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.findings-acl.1442.pdf
Checklist:
 2026.findings-acl.1442.checklist.pdf