Optimal Transport Posterior Alignment for Cross-lingual Semantic Parsing

TL;DR

This paper introduces a novel cross-lingual semantic parsing method that uses Optimal Transport to align latent variables, significantly improving performance in few-shot scenarios without relying on parallel translations.

Contribution

It proposes a new approach leveraging Optimal Transport for explicit divergence minimization in cross-lingual semantic parsing, achieving state-of-the-art results with limited data.

Findings

Outperforms previous methods on MTOP and MultiATIS++SQL datasets.

Improves performance even without parallel input translations.

Better captures cross-lingual structure in latent space.

Abstract

Cross-lingual semantic parsing transfers parsing capability from a high-resource language (e.g., English) to low-resource languages with scarce training data. Previous work has primarily considered silver-standard data augmentation or zero-shot methods, however, exploiting few-shot gold data is comparatively unexplored. We propose a new approach to cross-lingual semantic parsing by explicitly minimizing cross-lingual divergence between probabilistic latent variables using Optimal Transport. We demonstrate how this direct guidance improves parsing from natural languages using fewer examples and less training. We evaluate our method on two datasets, MTOP and MultiATIS++SQL, establishing state-of-the-art results under a few-shot cross-lingual regime. Ablation studies further reveal that our method improves performance even without parallel input translations. In addition, we show that our model better captures cross-lingual structure in the latent space to improve semantic representation similarity.