Learning Knowledge Bases with Parameters for Task-Oriented Dialogue Systems

Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2372–2394 November 16 - 20, 2020. c⃝2020 Association for Computational Linguistics 2372 Learning Knowledge Bases with Parameters for Task-Oriented Dialogue Systems Andrea Madotto, Samuel Cahyawijaya, Genta Indra Winata, Yan Xu, Zihan Liu, Zhaojiang Lin, Pascale Fung Center for Artificial Intelligence Research (CAiRE) Department of Electronic and Computer Engineering The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong {amadotto, scahyawijaya, giwinata, yxucb, zliucr, zlinao}@connect.ust.hk, pascale@ece.ust.hk Abstract Task-oriented dialogue systems are either mod- ularized with separate dialogue state tracking (DST) and management steps or end-to-end trainable. In either case, the knowledge base (KB) plays an essential role in fulfilling user requests. Modularized systems rely on DST to interact with the KB, which is expensive in terms of annotation and inference time. End- to-end systems use the KB directly as input, but they cannot scale when the KB is larger than a few hundred entries. In this paper, we propose a method to embed the KB, of any size, directly into the model parameters. The resulting model does not require any DST or template responses, nor the KB as input, and it can dynamically update its KB via fine- tuning. We evaluate our solution in five task- oriented dialogue datasets with small, medium, and large KB size. Our experiments show that end-to-end models can effectively em- bed knowledge bases in their parameters and achieve competitive performance in all evalu- ated datasets1. 1 Introduction Task-oriented dialogue systems are designed to help users achieve predefined goals, such as book- ing restaurants or movie recommendations via natu- ral language interactions. These systems are deeply connected with external Knowledge Bases (KBs) since the system responses are guided by the output from the KB and the dialogue history. The current state-of-the-arts (Lei et al., 2018; Zhang et al., 2019a; Mehri et al., 2019; Chen et al., 2019; Peng et al., 2020a; Hosseini-Asl et al., 2020) are end-to-end pipelined systems that rely on Dialogue State Tracking (DST) and Speech Act (S-ACT) annotations. Aside from the annotation cost, which is knowingly high (Budzianowski et al., 1Code available in https://github.com/ HLTCHKUST/ke-dialogue Figure 1: During training, the KE dialogues are gen- erated by fulfilling the TEMPLATE with the user goal query results, and they are used to embed the KB into the model parameter θ. At testing time, the model does not use any external knowledge to generate the correct responses. 2018), these pipelined systems must predict a valid DST for querying the KB, execute the query, gen- erate a response template, and finally fulfill it with the retrieved information. The resulting systems are usually overly complicated, and they require multiple steps, including a direct interaction with the KB. On the other end of the spectrum, there are end- to-end trainable models that use both the KB and the dialogue history as input, and they directly gen- erate system responses. Most of the implementa- tions use either the Gold KB as input (Eric et al., 2017a; Madotto et …

Content truncated to 500 words