Improving Zero-Shot Cross-Lingual Dialogue State Tracking via Contrastive Learning

Recent works in dialogue state tracking (DST) focus on a handful of languages, as collecting large-scale manually annotated data in different languages is expensive. Existing models address this issue by code-switched data augmentation or intermediate fine-tuning of multilingual pre-trained models. However, these models can only perform implicit alignment across languages. In this paper, we propose a novel model named Contrastive Learning for Cross-Lingual DST (CLCL-DST) to enhance zero-shot cross-lingual adaptation. Specifically, we use a self-built bilingual dictionary for lexical substitution to construct multilingual views of the same utterance. Then our approach leverages fine-grained contrastive learning to encourage representations of specific slot tokens in different views to be more similar than negative example pairs. By this means, CLCL-DST aligns similar words across languages into a more refined language-invariant space. In addition, CLCL-DST uses a significance-based keyword extraction approach to select task-related words to build the bilingual dictionary for better cross-lingual positive examples. Experiment results on Multilingual WoZ 2.0 and parallel MultiWoZ 2.1 datasets show that our proposed CLCL-DST outperforms existing state-of-the-art methods by a large margin, demonstrating the effectiveness of CLCL-DST.