基于数据选择和局部伪标注的跨语义依存分析研究(Selection and Pseudo Partial Annotationy)

Dazhan Mao (毛达展), Kuai Yu (喻快), Yanqiu Shao (邵艳秋)


Abstract
语义依存分析要走向实用,模型从单领域迁移到其他领域的领域适应能力至关重要。近年来,对抗学习针对领域适应这个任务取得了较好的效果,但对目标领域的无标注数据利用效率并不高。本文采用Self-training这种半监督学习方法,充分发挥无标注数据的潜能,弥补对抗学习方法的不足。但传统的Self-training效率和性能并不好,为此本文针对跨领域语义依存分析这个任务,尝试了强化学习数据选择器,提出了局部伪标注的标注策略,实验结果证明我们提出的模型优于基线模型。
Anthology ID:
2021.ccl-1.59
Volume:
Proceedings of the 20th Chinese National Conference on Computational Linguistics
Month:
August
Year:
2021
Address:
Huhhot, China
Editors:
Sheng Li (李生), Maosong Sun (孙茂松), Yang Liu (刘洋), Hua Wu (吴华), Kang Liu (刘康), Wanxiang Che (车万翔), Shizhu He (何世柱), Gaoqi Rao (饶高琦)
Venue:
CCL
SIG:
Publisher:
Chinese Information Processing Society of China
Note:
Pages:
655–666
Language:
Chinese
URL:
https://aclanthology.org/2021.ccl-1.59
DOI:
Bibkey:
Cite (ACL):
Dazhan Mao, Kuai Yu, and Yanqiu Shao. 2021. 基于数据选择和局部伪标注的跨语义依存分析研究(Selection and Pseudo Partial Annotationy). In Proceedings of the 20th Chinese National Conference on Computational Linguistics, pages 655–666, Huhhot, China. Chinese Information Processing Society of China.
Cite (Informal):
基于数据选择和局部伪标注的跨语义依存分析研究(Selection and Pseudo Partial Annotationy) (Mao et al., CCL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.ccl-1.59.pdf