Zero-Shot Dialogue State Tracking via Cross-Task Transfer

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 7890–7900 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 7890 Zero-Shot Dialogue State Tracking via Cross-Task Transfer Zhaojiang Lin1∗, Bing Liu2, Andrea Madotto1 ∗, Seungwhan Moon2, Paul Crook2, Zhenpeng Zhou2, Zhiguang Wang2, Zhou Yu3, Eunjoon Cho2, Rajen Subba2, Pascale Fung1 1The Hong Kong University of Science and Technology 2Facebook, 3Columbia University zlinao@ust.hk, bingl@fb.com Abstract Zero-shot transfer learning for dialogue state tracking (DST) enables us to handle a vari- ety of task-oriented dialogue domains without the expense of collecting in-domain data. In this work, we propose to transfer the cross- task knowledge from general question an- swering (QA) corpora for the zero-shot DST task. Specifically, we propose TransferQA, a transferable generative QA model that seam- lessly combines extractive QA and multi- choice QA via a text-to-text transformer frame- work, and tracks both categorical slots and non-categorical slots in DST. In addition, we introduce two effective ways to construct unan- swerable questions, namely, negative question sampling and context truncation, which enable our model to handle “none” value slots in the zero-shot DST setting. The extensive exper- iments show that our approaches substantially improve the existing zero-shot and few-shot re- sults on MultiWoz. Moreover, compared to the fully trained baseline on the Schema-Guided Dialogue dataset, our approach shows better generalization ability in unseen domains. 1 Introduction Virtual assistants are designed to help users per- form daily activities, such as travel planning, on- line shopping and restaurant booking. Dialogue state tracking (DST), as an essential component of these task-oriented dialogue systems, tracks users’ requirements throughout multi-turn conver- sations as dialogue states, which are typically in the form of a list of slot-value pairs. Training a DST model often requires extensive annotated di- alogue data. These data are often collected via a Wizard-of-Oz (Woz) (Kelley, 1984) setting, where two workers converse with each other and anno- tate the dialogue states of each utterance (Wen et al., 2017; Budzianowski et al., 2018; Moon et al., 2020), or with a Machines Talking To Machines ∗ Work done during internship at Facebook (M2M) framework (Shah et al., 2018), where di- alogues are synthesized via the system and user simulators (Campagna et al., 2020; Rastogi et al., 2020; Lin et al., 2021b). However, both of these approaches have inherent challenges when scaling to large datasets. For example, the data collec- tion process in a Woz setting incurs expensive and time-consuming manual annotations, while M2M requires exhaustive hand-crafted rules for covering various dialogue scenarios. In industrial applications, virtual assistants are required to add new services (domains) frequently based on user’s needs, but collecting extensive data for every new domain is costly and inefficient. Therefore, performing zero-shot prediction of di- alogue states is becoming increasingly important since it does not require the expense of data ac- quisition. There are mainly two lines of work in the zero-shot transfer learning problem. The first is cross-domain transfer learning (Wu et al., 2019; Kumar et al., 2020; Rastogi et al., 2020; …

Content truncated to 500 words