用细粒度标注提升多语言模型对网络暖心话的识别效果
AIxcellent Vibes at GermEval 2025 Shared Task on Candy Speech Detection: Improving Model Performance by Span-Level Training
- 采用片段级训练，让模型精准定位暖心话语
- XLM-RoBERTa-Large在双任务中分别取得0.8906和0.6307的高分
- 适合关注社交媒体正向内容检测的研究者参考
社交媒体中的积极支持性言论（暖心话）有助于促进网络文明，但其自动化检测仍缺乏系统研究，制约了对其影响的深入分析。本文基于46,000条德语YouTube评论数据，利用单语和多语言模型（包括GBERT、Qwen3 Embedding和XLM-RoBERTa）探究暖心话的可靠检测方法。结果表明，经过片段级训练的多语言XLM-RoBERTa-Large模型在GermEval 2025共享任务中表现最优，二分类任务F1达0.8906，分类片段检测任务严格F1为0.6307，位居榜首。研究推测，片段级训练策略、多语言能力及表情符号感知分词器共同提升了检测效果。实验验证了多语言模型在识别正向支持性语言方面的有效性。
原文摘要 · Abstract (English)
Positive, supportive online communication in social media (candy speech) has the potential to foster civility, yet automated detection of such language remains underexplored, limiting systematic analysis of its impact. We investigate how candy speech can be reliably detected in a 46k-comment German YouTube corpus by monolingual and multilingual language models, including GBERT, Qwen3 Embedding, and XLM-RoBERTa. We find that a multilingual XLM-RoBERTa-Large model trained to detect candy speech at the span level outperforms other approaches, ranking first in both binary positive F1: 0.8906) and categorized span-based detection (strict F1: 0.6307) subtasks at the GermEval 2025 Shared Task on Candy Speech Detection. We speculate that span-based training, multilingual capabilities, and emoji-aware tokenizers improved detection performance. Our results demonstrate the effectiveness of multilingual models in identifying positive, supportive language.
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。