@inproceedings{cao-etal-2022-sheng,
    title = "生成,推理与排序:基于多任务架构的数学文字题生成(Generating, Reasoning {\&} Ranking: Multitask Learning Framework for Math Word Problem Generation)",
    author = "Cao, Tianyang  and
      Xu, Xiaodan  and
      Chang, Baobao",
    editor = "Sun, Maosong  and
      Liu, Yang  and
      Che, Wanxiang  and
      Feng, Yang  and
      Qiu, Xipeng  and
      Rao, Gaoqi  and
      Chen, Yubo",
    booktitle = "Proceedings of the 21st Chinese National Conference on Computational Linguistics",
    month = oct,
    year = "2022",
    address = "Nanchang, China",
    publisher = "Chinese Information Processing Society of China",
    url = "https://preview.aclanthology.org/add-emnlp-2024-awards/2022.ccl-1.17/",
    pages = "178--189",
    language = "zho",
    abstract = "{\textquotedblleft}数学文字题是一段能反映数学等式潜在逻辑的叙述性文本。成功的数学问题生成在语言生成和教育领域都具有广阔的应用前景。前人的工作大多需要人工标注的模板或关键词作为输入,且未考虑数学表达式本身的特点。本文提出了一种多任务联合训练的问题文本生成模型。我们设计了三个辅助任务,包括数字间关系抽取、数值排序和片段替换预测。他们与生成目标联合训练,用以监督解码器的学习,增强模型对运算逻辑和问题条件的感知能力。实验证明所提方法能有效提升生成的数学文字题的质量。{\textquotedblright}"
}