基于生成式语言模型的立场检测探究(Research on Stance Detection with Generative Language Model)

Zhang Yuanshuo (张袁硕), Li Aohua (李澳华), Yin Zhaoning (尹召宁), Wang Panyi (王潘怡), Chen Bo (陈波), Zhao Xiaobing (赵小兵)


Abstract
“近年来,立场检测任务受到越来越多的关注,但相关标注数据在范围和规模上都有限,不能有效支撑基于神经网络的立场检测。为此,本文探索在零样本阯少样本场景下生成式语言模型在立场检测任务上的能力。首先,构建了一个全新的面向立场检测的数据集,包含5个主题,共2500个人工标注样例;然后,在此数据集上进行了一系列探索实验,实验结果表明:生成式语言模型在零样本设定下,采用结构化的提示学习表现良好;增加额外信息能够显著提升模型性能;在少样本设定下,提供相同目标的示例能够明显提升模型性能,而不同目标示例产生了负面作用;使用思维链可以显著提升模型性能;受提示学习的启发,微调预训练语言模型进一步论证提供额外信息对立场检测的增益显著。”
Anthology ID:
2024.ccl-1.37
Volume:
Proceedings of the 23rd Chinese National Conference on Computational Linguistics (Volume 1: Main Conference)
Month:
July
Year:
2024
Address:
Taiyuan, China
Editors:
Maosong Sun, Jiye Liang, Xianpei Han, Zhiyuan Liu, Yulan He
Venue:
CCL
SIG:
Publisher:
Chinese Information Processing Society of China
Note:
Pages:
481–491
Language:
Chinese
URL:
https://aclanthology.org/2024.ccl-1.37/
DOI:
Bibkey:
Cite (ACL):
Zhang Yuanshuo, Li Aohua, Yin Zhaoning, Wang Panyi, Chen Bo, and Zhao Xiaobing. 2024. 基于生成式语言模型的立场检测探究(Research on Stance Detection with Generative Language Model). In Proceedings of the 23rd Chinese National Conference on Computational Linguistics (Volume 1: Main Conference), pages 481–491, Taiyuan, China. Chinese Information Processing Society of China.
Cite (Informal):
基于生成式语言模型的立场检测探究(Research on Stance Detection with Generative Language Model) (Yuanshuo et al., CCL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.ccl-1.37.pdf