@inproceedings{yuchen-etal-2024-ufsc,
    title = "{UFSC}:基于统一特征空间构建的零样本关系抽取({UFSC}: A Unified Feature Space Construction for Zero-Shot Relation Extraction)",
    author = "Yuchen, Liu  and
      Jianyong, Duan  and
      Kang, Sun  and
      Qing, Zhang  and
      Li, He  and
      Hao, Wang  and
      Jie, Liu",
    editor = "Sun, Maosong  and
      Liang, Jiye  and
      Han, Xianpei  and
      Liu, Zhiyuan  and
      He, Yulan",
    booktitle = "Proceedings of the 23rd Chinese National Conference on Computational Linguistics (Volume 1: Main Conference)",
    month = jul,
    year = "2024",
    address = "Taiyuan, China",
    publisher = "Chinese Information Processing Society of China",
    url = "https://preview.aclanthology.org/jlcl-multiple-ingestion/2024.ccl-1.28/",
    pages = "370--381",
    language = "zho",
    abstract = "{\textquotedblleft}零样本关系抽取(ZSRE)旨在从可见关系中学习提取不可见关系的能力。一些研究表明:将样本语句与关系描述匹配进而预测不可见关系的方法,可以有效完成零样本关系抽取任务。然而,现有的匹配框架方法很少统一样本语句与关系描述的特征空间,缺乏对二者特征进行对齐。因此,本文提出一种为匹配框架零样本关系抽取而设计的统一特征空间构建方法。统一样本语句与关系描述的编码模块,并在此基础上引入特征相似损失。同时,为了减轻特征在空间上的聚合现象,引入特征均匀化模块,旨在构建特征更加均匀化的特征空间。本文所提出的方法实现了性能上的提升。与之前最佳的结果相比,在FewRel和Wiki-ZSL数据集上F1值平均提高1.6{\%}和3.4{\%},体现了统一特征空间构建以及特征均匀化模块的有效性。{\textquotedblright}"
}