Spider-KO: Korean Text-to-SQL Translation Dataset

Spider-KO is a Korean-language translation of the Yale University Spider dataset for text-to-SQL tasks. It contains natural language questions translated into Korean paired with their original English SQL queries across multiple domains. The dataset was created by huggingface-KREW and was last updated on 2025-05-29.

Use Cases

• Training Korean text-to-SQL models based on the translated natural language questions.
• Evaluating the performance of multilingual SQL generation systems on Korean inputs.
• Benchmarking cross-lingual transfer learning for semantic parsing tasks.
• Developing models that can handle SQL queries with English syntax from Korean prompts.

Strengths

• Based on the established Yale University Spider benchmark dataset.
• Provides both the original English questions and their Korean translations, enabling cross-lingual analysis.
• Covers multiple database domains as indicated in the description.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown.
• SQL queries remain in English, which may limit use cases requiring fully Korean SQL output.

Provenance

Source

Yale University Spider dataset, translated by huggingface-KREW.

Collection Method

Machine translation of natural language questions from English to Korean.

Freshness

Last updated 2025-05-29 05:30:36.