Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network
文献类型:期刊论文
作者 | Liu, Qing-Bin1,2; He, Shi-Zhu1,2; Liu, Cao3; Liu, Kang1,2; Zhao, Jun1,2 |
刊名 | JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY |
出版日期 | 2023-07-01 |
卷号 | 38期号:4页码:834-852 |
ISSN号 | 1000-9000 |
关键词 | end-to-end task-oriented dialogue dialogue state tracking (DST) unsupervised learning reinforcement learning |
DOI | 10.1007/s11390-021-1064-y |
通讯作者 | He, Shi-Zhu(shizhu.he@nlpr.ia.ac.cn) |
英文摘要 | This paper focuses on end-to-end task-oriented dialogue systems, which jointly handle dialogue state tracking (DST) and response generation. Traditional methods usually adopt a supervised paradigm to learn DST from a manually labeled corpus. However, the annotation of the corpus is costly, time-consuming, and cannot cover a wide range of domains in the real world. To solve this problem, we propose a multi-span prediction network (MSPN) that performs unsupervised DST for end-to-end task-oriented dialogue. Specifically, MSPN contains a novel split-merge copy mechanism that captures long-term dependencies in dialogues to automatically extract multiple text spans as keywords. Based on these keywords, MSPN uses a semantic distance based clustering approach to obtain the values of each slot. In addition, we propose an ontology-based reinforcement learning approach, which employs the values of each slot to train MSPN to generate relevant values. Experimental results on single-domain and multi-domain task-oriented dialogue datasets show that MSPN achieves state-of-the-art performance with significant improvements. Besides, we construct a new Chinese dialogue dataset MeDial in the low-resource medical domain, which further demonstrates the adaptability of MSPN. |
资助项目 | National Key Research and Development Program of China[2020AAA0106400] ; National Natural Science Foundation of China[61922085] ; National Natural Science Foundation of China[61976211] ; Independent Research Project of National Laboratory of Pattern Recognition[Z-2018013] ; Key Research Program of Chinese Academy of Sciences (CAS)[ZDBS-SSW-JSC006] ; Youth Innovation Promotion Association CAS[201912] |
WOS研究方向 | Computer Science |
语种 | 英语 |
出版者 | SPRINGER SINGAPORE PTE LTD |
WOS记录号 | WOS:001102032000008 |
资助机构 | National Key Research and Development Program of China ; National Natural Science Foundation of China ; Independent Research Project of National Laboratory of Pattern Recognition ; Key Research Program of Chinese Academy of Sciences (CAS) ; Youth Innovation Promotion Association CAS |
源URL | [http://ir.ia.ac.cn/handle/173211/55158] |
专题 | 复杂系统认知与决策实验室 |
通讯作者 | He, Shi-Zhu |
作者单位 | 1.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China 3.Beijing Sankuai Online Technol Co Ltd, Beijing 100102, Peoples R China |
推荐引用方式 GB/T 7714 | Liu, Qing-Bin,He, Shi-Zhu,Liu, Cao,et al. Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network[J]. JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY,2023,38(4):834-852. |
APA | Liu, Qing-Bin,He, Shi-Zhu,Liu, Cao,Liu, Kang,&Zhao, Jun.(2023).Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network.JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY,38(4),834-852. |
MLA | Liu, Qing-Bin,et al."Unsupervised Dialogue State Tracking for End-to-End Task-Oriented Dialogue with a Multi-Span Prediction Network".JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY 38.4(2023):834-852. |
入库方式: OAI收割
来源:自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。