RAST: Domain-Robust Dialogue Rewriting as Sequence Tagging

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 4913–4924 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 4913 RAST: Domain-Robust Dialogue Rewriting as Sequence Tagging Jie Hao1∗, Linfeng Song2†, Liwei Wang3∗, Kun Xu2, Zhaopeng Tu2, and Dong Yu2 1Florida State University, FL, USA haoj8711@gmail.com 2Tencent AI Lab, Bellevue, WA, USA {lfsong,kxkunxu,zptu,dyu}@tencent.com 3The Chinese University of Hong Kong lwwang@cse.cuhk.edu.hk Abstract The task of dialogue rewriting aims to recon- struct the latest dialogue utterance by copying the missing content from the dialogue context. Until now, the existing models for this task suffer from the robustness issue, i.e., perfor- mances drop dramatically when testing on a different dataset. We address this robustness issue by proposing a novel sequence-tagging- based model so that the search space is sig- nificantly reduced, yet the core of this task is still well covered. As a common issue of most tagging models for text generation, the model’s outputs may lack fluency. To allevi- ate this issue, we inject the loss signal from BLEU or GPT-2 under a REINFORCE frame- work. Experiments show huge improvements of our model over the current state-of-the-art systems when transferring to another dataset. 1 Introduction Recent years have witnessed increasing attention in conversation-based tasks, such as conversational question answering (Choi et al., 2018; Reddy et al., 2019; Sun et al., 2019), dialogue response genera- tion (Li et al., 2017; Zhang et al., 2018; Wu et al., 2019; Zhou et al., 2020), dialogue state tracking (Eric et al., 2020; Zeng et al., 2021) and dialogue understanding (Xu et al., 2021; Song et al., 2021; Yu et al., 2020), mainly due to increasing commer- cial demands. However, current models still face tremendous challenges in representing multi-turn dialogues, due to the frequent omission (a.k.a. el- lipsis) and coreference that people naturally use in conversations for brevity. Specifically, recent work (Su et al., 2019) has shown that ellipsis and coreference can exist in more than 70% of dialogue utterances. To tackle this problem, people have proposed coreference resolution and zero pronoun recovery. But, the state-of-the-art performances ∗Work done while J. Hao was interning and L. Wang was working at Tencent AI Lab. †Corresponding author. Turn Utterance with Translation u1 上海最近天气怎么样？ (How is the recent weather in Shanghai?) u2 最近经常阴天下雨。 (It is always raining recently.) u3 冬天就是这样。 (Winter is like this.) u′ 3 上海冬天就是经常阴天下雨。 (It is always raining in winter Shanghai.) Table 1: An example dialogue including the context utterances (u1 and u2), the latest utterance (u3) and the rewritten utterance (u′ 3). on these tasks are still far from satisfactory, not to mention their uncovered situations, such as when a whole verb phrase is omitted. Recently, the task of dialogue utterance rewriting (Su et al., 2019; Pan et al., 2019; Elgohary et al., 2019) was proposed as for explicitly representing multi-turn dialogues. The task aims to reconstruct the latest dialogue utterance into a new utterance that is semantically equivalent to the original one and can be understood without referring to the con- text. …

Content truncated to 500 words