More is Better: Enhancing Open-Domain Dialogue Generation via Multi-Source Heterogeneous Knowledge

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 2286–2300 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 2286 More is Better: Enhancing Open-Domain Dialogue Generation via Multi-Source Heterogeneous Knowledge Sixing Wu1, Ying Li4,5∗, Minghui Wang2, Dawei Zhang1, Yang Zhou3, Zhonghai Wu4,5 1School of Electronics Engineering and Computer Science, Peking University, Beijing, China 2School of Software and Microelectronics, Peking University, Beijing, China 3Auburn University, Auburn, Alabama, USA 4National Research Center of Software Engineering, Peking University, Beijing, China 5Key Lab of High Confidence Software Technologies (MOE), Peking University, Beijing, China Abstract Despite achieving remarkable performance, previous knowledge-enhanced works usually only use a single-source homogeneous knowl- edge base of limited knowledge coverage. Thus, they often degenerate into traditional methods because not all dialogues can be linked with knowledge entries. This paper proposes a novel dialogue generation model, MSKE-Dialog, to solve this issue with three unique advantages: (1) Rather than only one, MSKE-Dialog can simultaneously leverage multiple heterogeneous knowledge sources (it includes but is not limited to commonsense knowledge facts, text knowledge, infobox knowledge) to improve the knowledge cover- age; (2) To avoid the topic conflict among the context and different knowledge sources, we propose a Multi-Reference Selection to better select context/knowledge; (3) We pro- pose a Multi-Reference Generation to gener- ate informative responses by referring to mul- tiple generation references at the same time. Extensive evaluations on a Chinese dataset show the superior performance of this work against various state-of-the-art approaches. To our best knowledge, this work is the first to use the multi-source heterogeneous knowl- edge in the open-domain knowledge-enhanced dialogue generation. 1 Introduction The rapid developments of knowledge-enhanced techniques have enabled machines to understand the instinct semantics of human conversations fur- ther and generate informative responses (Yu et al., 2020). External knowledge, such as commonsense bases (Speer et al., 2017), documents (Zhao et al., 2020), and tables (Wu et al., 2021), can bridge the gap between machines and humans in conver- sation by generously providing knowledge that is ∗ Corresponding author: Ying Li, li.ying@pku.edu.cn. The email of the first author: wusixing@pku.edu.cn Commonsense (iPhone, related to, smartphone) (smartphone, has context ,mobile phones) (Android, has context, mobile phones) …. Text The iPhone is a line of smartphones designed and marketed by Apple Inc. that use Apple's iOS mobile operating system.… Infobox Table Entity Name: iPhone (Type: Smartphone) (Developer: Apple Inc.) (Services: iTunes, iCloud, Apple Music ) …. Dialogue Examples Query: What is your phone? Response: I am using an iPhone, because I love iOS and Apple Music. Figure 1: Examples of generating dialogues with multi- source heterogeneous knowledge. The number of knowledge sources can exceed one, and the knowledge structure can vary among sources. hard to be learned from a conversational corpus (Ghazvininejad et al., 2018). However, previous knowledge-enhanced works are still far from satisfactory because they usually solely rely on a single-source homogeneous knowl- edge base: (1) Conversations are diverse because humans are free to talk about whatever topics they like (Li et al., 2016; Hu et …

Content truncated to 500 words