Crossing the Conversational Chasm: A Primer on Natural Language Processing for Multilingual Task-Oriented Dialogue Systems

Abstract

Despite the fact that natural language conversations with machines represent one of the central objectives of AI, and despite the massive increase of research and development efforts in conversational AI, task-oriented dialogue (TOD) -i.e., conversations with an artificial agent with the aim of completing a concrete task -is currently limited to a few narrow domains (e.g., food ordering, ticket booking) and a handful of major languages (e.g., English, Chinese). In this work, we provide an extensive overview of existing efforts in multilingual TOD and analyse the factors preventing the development of truly multilingual TOD systems. We identify two main challenges that combined hinder the faster progress in multilingual TOD: (1) current state-of-the-art TOD models based on large pretrained neural language models are data hungry; at the same time (2) data acquisition for TOD use cases is expensive and tedious. Most existing approaches to multilingual TOD thus rely on (zero-or few-shot) cross-lingual transfer from resource-rich languages (in TOD, this is basically only English), either by means of (i) machine translation or (ii) multilingual representation spaces. However, such approaches are currently not a viable solution for a large number of low-resource languages without parallel data and/or limited monolingual corpora. Finally, we discuss critical challenges and potential solutions by drawing parallels between TOD and other cross-lingual and multilingual NLP research.

GlobalWoZ: Globalizing MultiWoZ to Develop Multilingual Task-Oriented Dialogue Systems

“…As noted by Razumovskaia et al (2021), there are two main designs for ToD systems: modular ToD systems and end-to-end ToD systems. In modular ToD systems, dialogue state tracking is an important component that parses the user's goal from the dialogue utterances (Wu et al, 2019b;Heck et al, 2020;Hosseini-Asl et al, 2020;Lin et al, 2020b).…”

“…However, most existing ToD systems are predominately build on English conversations, limiting their service for all of the world's citizens. The reason of this limitation lies in the stark lack of high-quality multilingual ToD datasets due to the high expense and challenges of human labeling (Razumovskaia et al, 2021).…”

Contextual Semantic Parsing for Multilingual Task-Oriented Dialogues

“…CrossWOZ (Zhu et al, 2020) and RiSAWOZ are Chinese datasets collected through crowdsourcing which is expensive and human errors degrade quality. The creation of affordable high-quality datasets for other languages still remains a challenge (Razumovskaia et al, 2021).…”

“…Underlying an effective TOD agent is dialogue state tracking, the task of predicting a formal representation of the conversation sufficient for the dialogue agent to reply, in the form of slots and values. However, DST currently remains restricted to a few popular languages (Razumovskaia et al, 2021). Traditional DST agents require large handannotated Wizard-of-Oz (Kelley, 1984) datasets for training, which are prohibitively labor-intensive to produce in most languages (Gunasekara et al, 2020).…”

BiToD: A Bilingual Multi-Domain Dataset For Task-Oriented Dialogue Modeling

“…However, existing datasets for end-to-end ToD modelling are limited to a single language, such as English [17,18], or Chinese [19,20]. The absence of bilingual or multilingual datasets not only limits the research on cross-lingual transfer learning [21] but also hinders the development of robust end-to-end ToD systems for multilingual countries and regions.…”