arXiv cs.MA· Micha{\l} Wawer, Jaros{\l}aw A. Chudziak·· 12 小时前AI 评分31
何时升级人工审核：面向不确定性感知内容审核的分歧模式
Knowing When to Escalate: Disagreement Patterns for Uncertainty-Aware Content Moderation
AI 导读
研究提出 Reasoning Trace Divergence Analysis（RTDA），对视角差异化智能体生成的推理依据做嵌入，按依据相似度与结论一致性将案例分为四类，从中识别出“推理相似但结论不同”的收敛性分歧，作为价值冲突的候选信号。
正文
Abstract:Agentic content-moderation systems increasingly rely on collective judgments from LLM agents, while aggregation methods commonly treat inter-agent disagreement as a reliability problem to be removed through consensus or voting. This framing is inappropriate for value-laden tasks, where disagreement may reflect irreducible value pluralism - a distinction standard uncertainty decompositions cannot recover, since they operate on output distributions and discard the reasoning behind each judgment. We propose Reasoning Trace Divergence Analysis (RTDA), which embeds model-generated justifications from perspective-differentiated agents and classifies each case by justification similarity and conclusion agreement. The resulting four-state taxonomy isolates convergent disagreement, where agents reasoning similarly but concluding differently, as a candidate signature of value-laden conflict. On 1,000 items from the Measuring Hate Speech corpus, this structural signal predicts human annotator disagreement better than magnitude-only and output-only baselines. We contribute a design principle: condition human escalation on the structure of disagreement, not its magnitude alone, with RTDA serving as a post-processing layer for systems that already generate reasoning traces.
| Comments: | Accepted at CIKM 2026, November 7-11, 2026, Rome, Italy |
| Subjects: | Multiagent Systems (cs.MA) |
| ACM classes: | I.2.11; I.2.7 |
| Cite as: | arXiv:2610.35335 [cs.MA] |
| (or arXiv:2610.35335v1 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35335 arXiv-issued DOI via DataCite (pending registration) |
|
| Related DOI: | https://doi.org/10.1145/3799682.3840059
DOI(s) linking to related resources |
Submission history
From: Michał Wawer [view email]
[v1]
Mon, 26 Oct 2026 12:32:25 UTC (204 KB)
来源：arXiv cs.MA · arxiv.org