arXiv cs.CL· Ghazal Kalhor, Sima Kamandloo, Mehrsa Shamsian, Behnam Bahrak·· 3 天前AI 评分22
从分诊到影响：面向可解释波斯语仇恨言论的多跳推理基准
From Triage to Impact: A Multi-Hop Reasoning Benchmark for Explainable Persian Hate Speech
AI 导读
研究者推出一个细粒度多跳基准，用于可解释的波斯语仇恨言论分析，覆盖内容分诊、目标识别、仇恨策略与道德基础分析、文本证据抽取、有害叙事推断和严重程度评估。评测显示，任务越依赖细粒度语义与推理，主流 LLM 表现越差；闭源模型整体优于开放权重模型，推理型模型并无稳定优势。引入中间推理步骤相比一次性评估能显著提升严重程度预测，基准已公开。
正文 · 原文
Abstract:Hate speech has become increasingly prevalent on social media, raising concerns about its detection, interpretation, and moderation. Persian, a relatively low-resource language, lacks sufficient annotated data and computational resources for fine-grained hate speech analysis. In this study, we introduce a fine-grained, multi-hop benchmark for explainable Persian hate speech analysis, covering content triage, target identification, hate tactic and moral foundation analysis, textual evidence extraction, harmful narrative inference, and severity assessment. By evaluating state-of-the-art large language models (LLMs), we find that performance generally declines as tasks require more fine-grained semantic and inferential reasoning. Proprietary models generally outperform open-weight models, while reasoning-oriented models show no consistent advantage. Incorporating intermediate reasoning steps substantially improves severity prediction compared with one-shot assessment. Our analysis further reveals that Incitement to Violence or Harm is most strongly associated with higher severity, while political-affiliation targets are particularly associated with subversion-based moral framing. Overall, this work provides insights into LLM-based hate speech analysis and supports more accountable moderation of Persian-language social media. The benchmark is publicly available at this https URL.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.20370 [cs.CL] |
| (or arXiv:2610.20370v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20370 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ghazal Kalhor [view email]
[v1]
Wed, 14 Oct 2026 15:42:15 UTC (1,120 KB)
来源：arXiv cs.CL · arxiv.org