arXiv cs.CL· Ivana Crescenzi, Arianna Muti, Maarten Sap, Tommaso Caselli, Debora Nozza·· 12 小时前AI 评分20
用 RAG 对抗厌女：提升隐晦厌女解释质量的研究
RAGing Against Misogyny: Improving Implicit Misogyny Explanations
AI 导读
研究提出一个 RAG 框架，用于检测和解释英语与意大利语中的隐晦厌女内容，知识源包括百科、俚语和女性主义文献。在 ImplicIT-Mis 上 RAG 提升了召回率，但在 SBIC+ 上未改善检测表现；人工评估显示两种语言中 RAG 生成的解释均更受偏好，女性主义文献是最稳定的有效知识源。
正文
Abstract:Implicit misogyny often requires background knowledge to be understood. To address this challenge, we propose a Retrieval-Augmented Generation (RAG) framework for detecting and explaining implicit misogyny in English and Italian, leveraging encyclopedic, slang and feminist knowledge sources. Using the ImplicIT-Mis and SBIC+ datasets, we evaluate the contribution of each knowledge source to detection and explanation quality, investigating also whether RAG addresses known LLM failure modes in implicit misogyny. Results show that RAG improves recall on ImplicIT-Mis, but does not improve detection performance on SBIC+. However, human evaluation shows that RAG-based explanations are preferred over no-RAG explanations in both languages. Feminist literature emerges as the most consistently useful source for detection and explanation.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.35094 [cs.CL] |
| (or arXiv:2610.35094v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35094 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ivana Crescenzi [view email]
[v1]
Mon, 26 Oct 2026 09:35:01 UTC (11,729 KB)
来源：arXiv cs.CL · arxiv.org