用融合BERT模型检测印地语和尼泊尔语的仇恨言论，提升识别准确率。
NLPineers@ NLU of Devanagari Script Languages 2025: Hate Speech Detection using Ensembling of BERT-based models
- 融合多语言BERT模型，提升对印地语与尼泊尔语仇恨言论的识别能力。
- 在数据不平衡情况下，通过回译增强并保持标签一致性，召回率达0.7762。
- 为南亚语言仇恨言论检测提供可复用框架，适合相关研究者参考。
本文针对CHIPSAL@COLING 2025共享任务的子任务B，研究使用XLM-RoBERTa、MURIL和IndicBERT等基于Transformer的模型，在印地语和尼泊尔语等德文阿加里文字母语言中进行仇恨言论检测。面对仇恨言论与言论自由之间的微妙边界，提出融合多语言BERT模型的方法，有效提升了检测性能。最佳模型在测试集上达到0.7762的召回率（排名3/31）和0.6914的F1分数（排名17/31）。为缓解类别不平衡问题，采用回译进行数据增强，并利用余弦相似度确保增强后标签的一致性。本工作强调了在德文阿加里文字母语言中开展仇恨言论检测的重要性，并为后续研究奠定了基础。
原文摘要 · Abstract (English)
This paper explores hate speech detection in Devanagari-scripted languages, focusing on Hindi and Nepali, for Subtask B of the CHIPSAL@COLING 2025 Shared Task. Using a range of transformer-based models such as XLM-RoBERTa, MURIL, and IndicBERT, we examine their effectiveness in navigating the nuanced boundary between hate speech and free expression. Our best performing model, implemented as ensemble of multilingual BERT models achieve Recall of 0.7762 (Rank 3/31 in terms of recall) and F1 score of 0.6914 (Rank 17/31). To address class imbalance, we used backtranslation for data augmentation, and cosine similarity to preserve label consistency after augmentation. This work emphasizes the need for hate speech detection in Devanagari-scripted languages and presents a foundation for further research.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。