Prompter: Zero-shot Adaptive Prefixes for Dialogue State Tracking Domain Adaptation

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 4588–4603 July 9-14, 2023 ©2023 Association for Computational Linguistics Prompter: Zero-shot Adaptive Prefixes for Dialogue State Tracking Domain Adaptation Taha Aksu†‡∗,, Min-Yen Kan†, Nancy F. Chen‡ † National University of Singapore (NUS), Singapore ‡ Institute for Infocomm Research (I2R), A*STAR, Singapore *taksu@u.nus.edu Abstract A challenge in the Dialogue State Tracking (DST) field is adapting models to new domains without using any supervised data — zero-shot domain adaptation. Parameter-Efficient Trans- fer Learning (PETL) has the potential to ad- dress this problem due to its robustness. How- ever, it has yet to be applied to the zero-shot scenarios, as it is not clear how to apply it un- supervisedly. Our method, Prompter, uses descriptions of target domain slots to generate dynamic pre- fixes that are concatenated to the key and val- ues at each layer’s self-attention mechanism. This allows for the use of prefix-tuning in zero- shot. Prompter outperforms previous methods on both the MultiWOZ and SGD benchmarks. In generating prefixes, our analyses find that Prompter not only utilizes the semantics of slot descriptions but also how often the slots appear together in conversation. Moreover, Prompter’s gains are due to its improved ability to distin- guish “none”-valued dialogue slots, compared against baselines. 1 Introduction Task-oriented dialogue (TOD) systems serve users through several tasks, such as booking a table in a restaurant or suggesting tourist attractions. One crucial component of these systems, Dialogue State Tracking (DST), is responsible for extracting users’ preferences (i.e. slot-values) over key attributes (i.e. slot-labels) of their service (Wu et al., 2019). DST has a significant role in TOD systems as it ensures that both the action taken in the back-end and the responses returned to the users are aligned with the preferences that the users indicate. A challenging task in this field is to adapt an existing DST model to a new domain it has not seen before without using any supervised data, i.e. in the zero-shot scenario. This is important, as in many new scenarios, it is hard to collect data, let alone annotate it. Yet it is still an essential need Figure 1: Zero-shot domain adaptation. The model is trained on four source domains and tested on the train-booking domain without any supervised training. Bottom-left: T5 baseline predictions, Bottom-right: Prompter predictions. (Correct, incorrect) predictions are colored (green, red), respectively. for a TOD system to appropriately answer such queries in new contexts. The challenge arises from the differences in dialogue context, slot values, and slot labels among different domains. For example, a model could be trained on the ‘taxi-booking’ do- main and thus capable of extracting the destination for a taxi; but when deployed to the ‘train-booking’ domain, the range of slot-values changes, resulting in a higher probability of a mistaken inference. We show an example (Figure 1), where due to the su- perficial connections a baseline T5 model forms, it 4588 incorrectly predicts ‘Ashley Hotel’ as the train des- …

Content truncated to 500 words