用合成数据解决多语言多标签情感分类数据少的问题
Multilingual Multi-Label Emotion Classification at Scale with Synthetic Data
- 构建100万条跨23语言的合成情感数据，每语言5万条
- XLM-R-Large模型在自建测试集上达0.868 F1-micro
- 零样本迁移下超越专用英文模型，支持全部23语言
多语言情感分类受限于标注数据稀缺：现有语料库以英语为主，单标签，覆盖语言有限。本文构建了一个大规模合成训练语料库，包含超过100万条多标签样本（每语言5万条），涵盖23种语言（阿拉伯语、孟加拉语、荷兰语、英语、法语、德语、印地语、印尼语、意大利语、日语、韩语、中文、波兰语、葡萄牙语、旁遮普语、俄语、西班牙语、斯瓦希里语、泰米尔语、土耳其语、乌克兰语、乌尔都语、越南语），使用文化适配生成和程序化质量过滤，覆盖11种情绪类别。在相同条件下训练并比较六种多语言Transformer编码器，从DistilBERT（1.35亿参数）到XLM-R-Large（5.6亿参数）。在自建领域内测试集上，XLM-R-Large取得0.868 F1-micro和0.987 AUC-micro。为验证效果，所有模型在零样本下评估于GoEmotions（英语）和SemEval-2018 Task 1 E-c（英语、阿拉伯语、西班牙语）。在无阈值排名指标上，XLM-R-Large匹配或超越仅英语专用模型，AP-micro（0.636）和LRAP（0.804）持平，AUC-micro（0.810 vs. 0.787）更优，且原生支持全部23种语言。最佳基础模型已公开发布于https://huggingface.co/tabularisai/multilingual-emotion-classification。
原文摘要 · Abstract (English)
Emotion classification in multilingual settings remains constrained by the scarcity of annotated data: existing corpora are predominantly English, single-label, and cover few languages. We address this gap by constructing a large-scale synthetic training corpus of over 1M multi-label samples (50k per language) across 23 languages: Arabic, Bengali, Dutch, English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Mandarin, Polish, Portuguese, Punjabi, Russian, Spanish, Swahili, Tamil, Turkish, Ukrainian, Urdu, and Vietnamese, covering 11 emotion categories using culturally-adapted generation and programmatic quality filtering. We train and compare six multilingual transformer encoders, from DistilBERT (135M parameters) to XLM-R-Large (560M parameters), under identical conditions. On our in-domain test set, XLM-R-Large achieves 0.868 F1-micro and 0.987 AUC-micro. To validate against human-annotated data, we evaluate all models zero-shot on GoEmotions (English) and SemEval-2018 Task 1 E-c (English, Arabic, Spanish). On threshold-free ranking metrics, XLM-R-Large matches or exceeds English-only specialist models, tying on AP-micro (0.636) and LRAP (0.804) while surpassing on AUC-micro (0.810 vs. 0.787), while natively supporting all 23 languages. The best base-sized model is publicly available at https://huggingface.co/tabularisai/multilingual-emotion-classification
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。