Research map / Large language models / Language, safety and interpretability
MultiSpider: Towards Benchmarking Multilingual Text-to-SQL Semantic Parsing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add SAVe schema augmentation to mBART | mBART · MultiSpider, monolingual training, AVG (6 langs) | exact-match accuracy: 43.6 -> 45.5 (+1.9%) | Helped |
| Add SAVe schema augmentation to RAT-SQL + XLM-R | RAT-SQL + XLM-R · MultiSpider, monolingual training, AVG (6 langs) | exact-match accuracy: 61.8 -> 63.2 (+1.4%) | Helped |
| Add SAVe schema augmentation to mBART | mBART · MultiSpider, multilingual training, AVG (6 langs) | exact-match accuracy: 47.5 -> 49.8 (+2.3%) | Helped |
| Add SAVe schema augmentation to RAT-SQL + XLM-R | RAT-SQL + XLM-R · MultiSpider, multilingual training, AVG (6 langs) | exact-match accuracy: 65.2 -> 66.7 (+1.5%) | Helped |
| Use multilingual training data instead of target-language-only training | mBART and RAT-SQL + XLM-R · MultiSpider | exact-match accuracy: +3.9% | Helped |
| Translate target-language question and schema into English at inference instead of directly predicting | Zero-shot text-to-SQL parsers · MultiSpider, zero-shot transfer | exact-match accuracy: about 1.6% | Hurt |
| Verify generated schema candidates with NLI | SAVe augmented schemas · Human evaluation of augmented schemas | semantic-equivalence accuracy: 33.2% -> 74.5% | Helped |
| Use SAVe for English rather than non-English inputs | SAVe · MultiSpider, monolingual training | exact-match accuracy: 0.2%-1.0%; 1.4%-1.9% | Mixed |
Closest papers by meaning
- ROUTE: Robust Multitask Tuning and Collaboration for Text-to-SQL (ICLR 2025 · similarity 0.87)
- SQLens: An End-to-End Framework for Error Detection and Correction in Text-to-SQL (NeurIPS 2025 · similarity 0.87)
- MDPBench: A Benchmark for Multilingual Document Parsing in Real-World Scenarios (NeurIPS 2026 · similarity 0.87)
- Structure-Guided Large Language Models for Text-to-SQL Generation (ICML 2025 · similarity 0.86)
- PARROT: A Benchmark for Evaluating LLMs in Cross-System SQL Translation (NeurIPS 2025 · similarity 0.86)
- SQLFixAgent: Towards Semantic-Accurate Text-to-SQL Parsing via Consistency-Enhanced Multi-Agent Collaboration (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.