Zero-Shot Cross-lingual Semantic Parsing
This addresses the problem of building semantic parsers for new languages without costly parallel data, which is incremental as it builds on existing cross-lingual parsing work but removes key assumptions.
The paper tackles cross-lingual semantic parsing without parallel data or machine translation tools by proposing a multi-task encoder-decoder model that uses only English-logical form pairs and in-domain corpora in new languages, achieving performance significantly above translation-based baselines and sometimes competing with supervised upper-bounds.
Recent work in cross-lingual semantic parsing has successfully applied machine translation to localize parsers to new languages. However, these advances assume access to high-quality machine translation systems and word alignment tools. We remove these assumptions and study cross-lingual semantic parsing as a zero-shot problem, without parallel data (i.e., utterance-logical form pairs) for new languages. We propose a multi-task encoder-decoder model to transfer parsing knowledge to additional languages using only English-logical form paired data and in-domain natural language corpora in each new language. Our model encourages language-agnostic encodings by jointly optimizing for logical-form generation with auxiliary objectives designed for cross-lingual latent representation alignment. Our parser performs significantly above translation-based baselines and, in some cases, competes with the supervised upper-bound.