Zero-shot language extension for dialogue state tracking via pre-trained models and multi-auxiliary-tasks fine-tuning

doi:10.1016/j.knosys.2022.110015

	Zero-shot language extension for dialogue state tracking via pre-trained models and multi-auxiliary-tasks fine-tuning
	Xiang, Lu1,2 ; Zhao, Yang1,2 ; Zhu, Junnan1,2 ; Zhou, Yu1,2,3 ; Zong, Chengqing1,2
发表期刊	KNOWLEDGE-BASED SYSTEMS
ISSN	0950-7051
	2023-01-10
卷号	259 页码:14
通讯作者	Zhou, Yu(yzhou@nlpr.ia.ac.cn)
摘要	Dialogue state tracking (DST), a crucial component of the task-oriented dialogue system (TOD), is designed to track the user's goal. Existing DST models mainly focus on monolingual dialogue input, failing to meet the growing needs of a TOD to provide multilingual services. Therefore, this paper proposes a novel Zero-shot Language Extension scenario for DST, extending the monolingual DST to multilingual DST without extra high-cost dialogue data annotation. In this scenario, the multilingual DST only needs a single shared model to handle multilingual input and generate a unified dialogue state. This setting makes deploying a complete multilingual TOD easy since it could be reused by the downstream components from existing monolingual TOD. Specifically, we achieve the language extension by multi-auxiliary-tasks fine-tuning of multilingual pre-trained models, where five relevant auxiliary tasks are jointly designed, including monolingual DST, cross-lingual DST, forward word translation, utterance recovery, and semantic similarity. The extended multilingual DST model can be enhanced through joint optimization with all the auxiliary tasks by capturing multilingual context understanding and cross-lingual alignment characteristics. Comprehensive experiments on Multilingual WOZ dataset (English -> German and English -> Italian) and cross-lingual MultiWOZ dataset (English -> Chinese and Chinese -> English) demonstrate the effectiveness and superiority of the proposed method.(c) 2022 Elsevier B.V. All rights reserved.
关键词	Dialogue state tracking Zero -shot language extension Multilingual DST Pre -trained models Multi -auxiliary -tasks fine-tuning
DOI	10.1016/j.knosys.2022.110015
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China ; [2020AAA0108600]
项目资助者	National Key R&D Program of China
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:000883002100001
出版者	ELSEVIER
引用统计	被引频次：2[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/51249
专题	多模态人工智能系统全国重点实验室
通讯作者	Zhou, Yu
作者单位	1.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing, Peoples R China 3.Zhongke Fanyu Technol Co Ltd, Fanyu AI Lab, Beijing, Peoples R China
第一作者单位	模式识别国家重点实验室
通讯作者单位	模式识别国家重点实验室
推荐引用方式 GB/T 7714	Xiang, Lu,Zhao, Yang,Zhu, Junnan,et al. Zero-shot language extension for dialogue state tracking via pre-trained models and multi-auxiliary-tasks fine-tuning[J]. KNOWLEDGE-BASED SYSTEMS,2023,259:14.
APA	Xiang, Lu,Zhao, Yang,Zhu, Junnan,Zhou, Yu,&Zong, Chengqing.(2023).Zero-shot language extension for dialogue state tracking via pre-trained models and multi-auxiliary-tasks fine-tuning.KNOWLEDGE-BASED SYSTEMS,259,14.
MLA	Xiang, Lu,et al."Zero-shot language extension for dialogue state tracking via pre-trained models and multi-auxiliary-tasks fine-tuning".KNOWLEDGE-BASED SYSTEMS 259(2023):14.