Cross-lingual Transfer Learning for Low-Resource Semantic Parsing

You R. Name, David Lee, Emma Clark

Advances in Neural Information Processing Systems (NeurIPS), 2025

abstract

Semantic parsing in low-resource languages remains a significant challenge due to the scarcity of annotated data. We propose a cross-lingual transfer framework that leverages multilingual pre-trained models and language-agnostic meaning representations to enable effective transfer from high-resource to low-resource languages. Our approach introduces a novel alignment objective that maps syntactically diverse languages into a shared semantic space. Experiments across 8 languages on two semantic parsing benchmarks demonstrate that our method achieves competitive performance using only 10% of the target language training data.

@inproceedings{name2025crosslingual,
  title={Cross-lingual Transfer Learning for Low-Resource Semantic Parsing},
  author={Name, You R. and Lee, David and Clark, Emma},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
  year={2025}
}