@inproceedings{rao-etal-2023-ji,
    title = "基于{R}o{BERT}a的中文仇恨言论侦测方法研究({C}hinese Hate Speech detection method Based on {R}o{BERT}a-{WWM})",
    author = "Rao, Xiaojun  and
      Zhang, Yangsen  and
      Jia, Qilong  and
      Liu, Xueyang",
    editor = "Sun, Maosong  and
      Qin, Bing  and
      Qiu, Xipeng  and
      Jiang, Jing  and
      Han, Xianpei",
    booktitle = "Proceedings of the 22nd Chinese National Conference on Computational Linguistics",
    month = aug,
    year = "2023",
    address = "Harbin, China",
    publisher = "Chinese Information Processing Society of China",
    url = "https://preview.aclanthology.org/sigarab-more-entries-6621/2023.ccl-1.44/",
    pages = "501--511",
    language = "zho",
    abstract = "``随着互联网的普及,社交媒体虽然提供了交流观点的平台,但因其虚拟性和匿名性也加剧了仇恨言论的传播,因此自动侦测仇恨言论对于维护社交媒体平台的文明发展至关重要。针对以上问题,构建了一个中文仇恨言论数据集CHSD,并提出了一种中文仇恨言论侦测模型RoBERTa-CHHSD。该模型首先采用RoBERTa预训练语言模型对中文仇恨言论进行序列化处理,提取文本特征信息;再分别接入TextCNN模型和Bi-GRU模型,提取多层次局部语义特征和句子间全局依赖关系信息;将二者结果融合来提取文本中更深层次的仇恨言论特征,对中文仇恨言论进行分类,从而实现中文仇恨言论的侦测。实验结果表明,本模型在CHSD数据集上的F1值为89.12{\%},与当前最优主流模型RoBERTa-WWM相比提升了1.76{\%}。''"
}