Smoothing Dialogue States for Open Conversational Machine Reading

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 3685–3696 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 3685 Smoothing Dialogue States for Open Conversational Machine Reading Zhuosheng Zhang1,2,3,#, Siru Ouyang1,2,3,#, Hai Zhao1,2,3,∗, Masao Utiyama4, Eiichiro Sumita4 1 Department of Computer Science and Engineering, Shanghai Jiao Tong University 2 Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering, Shanghai Jiao Tong University 3MoE Key Lab of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University 4National Institute of Information and Communications Technology (NICT), Kyoto, Japan {zhangzs,oysr0926}@sjtu.edu.cn,zhaohai@cs.sjtu.edu.cn {mutiyama,eiichiro.sumita}@nict.go.jp Abstract Conversational machine reading (CMR) re- quires machines to communicate with humans through multi-turn interactions between two salient dialogue states of decision making and question generation processes. In open CMR settings, as the more realistic scenario, the re- trieved background knowledge would be noisy, which results in severe challenges in the in- formation transmission. Existing studies com- monly train independent or pipeline systems for the two subtasks. However, those methods are trivial by using hard-label decisions to acti- vate question generation, which eventually hin- ders the model performance. In this work, we propose an effective gating strategy by smooth- ing the two dialogue states in only one decoder and bridge decision making and question gen- eration to provide a richer dialogue state refer- ence. Experiments on the OR-ShARC dataset show the effectiveness of our method, which achieves new state-of-the-art results. 1 Introduction The ultimate goal of multi-turn dialogue is to en- able the machine to interact with human beings and solve practical problems (Zhu et al., 2018; Zhang et al., 2018; Zaib et al., 2020; Huang et al., 2020; Fan et al., 2020; Gu et al., 2021). It usually adopts the form of question answering (QA) according to the user’s query along with the dialogue context (Sun et al., 2019; Reddy et al., 2019; Choi et al., 2018). The machine may also actively ask ques- tions for confirmation (Wu et al., 2018; Cai et al., 2019; Zhang et al., 2020b; Gu et al., 2020). In the classic spoken language understanding tasks (Tur and De Mori, 2011; Zhang et al., 2020a; Ren et al., 2018; Qin et al., 2021), specific slots and intentions are usually defined. According to these predefined patterns, the machine interacts ∗Corresponding author. # Equal contribution. This paper was partially supported by Key Projects of National Natural Science Foundation of China (U1836222 and 61733011). with people according to the dialogue states, and completes specific tasks, such as ordering meals (Liu et al., 2013) and air tickets (Price, 1990). In real-world scenario, annotating data such as in- tents and slots is expensive. Inspired by the studies of reading comprehension (Rajpurkar et al., 2016, 2018; Zhang et al., 2020c, 2021), there appears a more general task — conversational machine read- ing (CMR) (Saeidi et al., 2018): given the inquiry, the machine is required to retrieve relevant sup- porting rule documents, the machine should judge whether the goal is satisfied according to the dia- logue context, and make …

Content truncated to 500 words