KdConv: A Chinese Multi-domain Dialogue Dataset Towards Multi-turn Knowledge-driven Conversation

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7098–7108 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 7098 KdConv: A Chinese Multi-domain Dialogue Dataset Towards Multi-turn Knowledge-driven Conversation Hao Zhou∗, Chujie Zheng∗, Kaili Huang, Minlie Huang†, Xiaoyan Zhu Department of Computer Science and Technology, Institute for Artificial Intelligence, State Key Lab of Intelligent Technology and Systems, Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing 100084, China {tuxchow,chujiezhengchn}@gmail.com, aihuang@tsinghua.edu.cn Abstract The research of knowledge-driven conversa- tional systems is largely limited due to the lack of dialog data which consists of multi-turn con- versations on multiple topics and with knowl- edge annotations. In this paper, we propose a Chinese multi-domain knowledge-driven con- versation dataset, KdConv, which grounds the topics in multi-turn conversations to knowl- edge graphs. Our corpus contains 4.5K conver- sations from three domains (film, music, and travel), and 86K utterances with an average turn number of 19.0. These conversations con- tain in-depth discussions on related topics and natural transition between multiple topics. To facilitate the following research on this corpus, we provide several benchmark models. Com- parative results show that the models can be enhanced by introducing background knowl- edge, yet there is still a large space for lever- aging knowledge to model multi-turn conver- sations for further research. Results also show that there are obvious performance differences between different domains, indicating that it is worth further explore transfer learning and do- main adaptation. The corpus and benchmark models are publicly available1. 1 Introduction It has been a long-term goal of artificial intelli- gence to deliver human-like conversations, where background knowledge plays a crucial role in the success of conversational systems (Shang et al., 2015; Li et al., 2016a; Shao et al., 2017). In task- oriented dialog systems, background knowledge is defined as slot-value pairs, which provides key information for question answering or recommen- dation, and has been well defined and thoroughly studied (Wen et al., 2015; Zhou et al., 2016). In ∗ Equal contribution † Corresponding author: Minlie Huang. 1https://github.com/thu-coai/KdConv open-domain conversational systems, it is impor- tant but challenging to leverage background knowl- edge, which is represented as either knowledge graphs (Zhu et al., 2017; Zhou et al., 2018a) or unstructured texts (Ghazvininejad et al., 2018), for making effective interactions. Recently, a variety of knowledge-grounded con- versation corpora have been proposed (Zhou et al., 2018b; Dinan et al., 2018; Moghe et al., 2018; Moon et al., 2019; Wu et al., 2019; Liu et al., 2018; Tuan et al., 2019; Qin et al., 2019) to fill the gap where previous datasets do not provide knowledge grounding of the conversations (God- frey et al., 1992; Shang et al., 2015; Lowe et al., 2015). CMU DoG (Zhou et al., 2018b), India DoG (Moghe et al., 2018), and Wizard of Wikipedia (Dinan et al., 2018) demonstrate attempts for gen- erating informative responses with topic-related Wikipedia articles. However, these datasets are not suitable for modeling topic transition or knowl- edge planning through multi-turn …

Content truncated to 500 words