Abstract
"意图识别与槽位填充是口语理解中的两个子任务,联合建模这两项任务能够利用共享特征提升任务间的协同建模效果。然而,现有方法普遍缺乏对句子主题语义的显式建模,难以捕捉更充分的全局语义信息,尤其在多意图场景下系统建模性能下降严重。为缓解上述问题,本文提出了一种主题感知的意图识别与槽位填充联合建模方法,该方法构造了主题提取模块以学习句子主题分布表示,结合主题引导的意图和槽位表示增强网络插入主题信息,使得模型在识别句子意图和填充槽位过程中能够显式建模主题信息。实验结果表明,本文所提出方法在多意图公开数据集MixATIS和MixSNIPS上分别获得了50.9%和84.8%的整体准确率,相较多个基线模型取得了更优的性能表现。"
- Anthology ID:
- 2025.ccl-1.39
- Volume:
- Proceedings of the 24th China National Conference on Computational Linguistics (CCL 2025)
- Month:
- August
- Year:
- 2025
- Address:
- Jinan, China
- Editors:
- Maosong Sun, Peiyong Duan, Zhiyuan Liu, Ruifeng Xu, Weiwei Sun
- Venue:
- CCL
- SIG:
- Publisher:
- Chinese Information Processing Society of China
- Note:
- Pages:
- 525–536
- Language:
- URL:
- https://aclanthology.org/2025.ccl-1.39/
- DOI:
- Bibkey:
- Cite (ACL):
- Jing Luo, Weihua Wang, Yue Cao, Feilong Bao, and Guanglai Gao. 2025. 主题感知的多意图识别与槽位填充联合建模方法. In Proceedings of the 24th China National Conference on Computational Linguistics (CCL 2025), pages 525–536, Jinan, China. Chinese Information Processing Society of China.
- Cite (Informal):
- 主题感知的多意图识别与槽位填充联合建模方法 (Luo et al., CCL 2025)
- Copy Citation:
- PDF:
- https://aclanthology.org/2025.ccl-1.39.pdf