This dataset is a merged collection of multiple text-to-SQL datasets, designed to provide a comprehensive resource for training and evaluating text-to-SQL models. It combines data from several popular benchmarks, including Spider, CoSQL, SparC, and others, to create a diverse and robust dataset for natural language to SQL query generation tasks.
Dataset Details Dataset Description Curated by: Mudasir Ahmad Mir
Language(s) (NLP): English
License: Apache 2.0
This dataset is ideal for researchers and developers working on natural language processing (NLP), semantic parsing, and database query generation. It supports a wide range of SQL complexities, from simple queries to nested and multi-turn interactions, making it suitable for both beginner and advanced text-to-SQL tasks.
Dataset Sources [optional] Repository: Hugging Face Dataset Link
Original Datasets:
Spider: Yale-LILY Lab
CoSQL: Yale-LILY Lab
SparC: Yale-LILY Lab
Uses Direct Use This dataset is intended for:
Training Text-to-SQL Models: Use this dataset to train models for converting natural language questions into SQL queries.
Benchmarking: Evaluate the performance of text-to-SQL models across diverse queries and domains.
Research: Study the challenges of semantic parsing, cross-domain generalization, and conversational SQL generation.
Out-of-Scope Use This dataset is not suitable for:
Tasks requiring domain-specific knowledge beyond the included datasets.
Applications requiring real-time or low-latency SQL generation without further fine-tuning.
Dataset Structure The dataset contains the following columns:
sentence: A natural language question or query.
sql: The corresponding SQL query for the given sentence.
Dataset Composition The dataset is created by merging the following sources:
Spider Dataset: A cross-domain text-to-SQL dataset with complex SQL queries.
CoSQL Dataset: A conversational text-to-SQL dataset with multi-turn interactions.
SparC Dataset: A context-dependent text-to-SQL dataset for cross-domain tasks.
Custom Text-to-SQL Dataset: A dataset containing additional sentence-SQL pairs.
Train Dataset: A dataset with context-question-answer pairs, adapted for text-to-SQL tasks.
Dataset Statistics Total Rows: [Insert total number of rows]
Unique SQL Patterns: [Insert number of unique SQL patterns]
Domains Covered: Academia, geography, conversational systems, and more.
Dataset Creation Curation Rationale This dataset was created to provide a unified and diverse resource for text-to-SQL tasks, combining high-quality datasets from multiple domains and use cases. The goal is to support research and development in natural language understanding and database query generation.
Source Data Data Collection and Processing The dataset was created by merging publicly available text-to-SQL datasets, including Spider, CoSQL, and SparC. The data was cleaned and standardized to ensure consistency in column names and formats.
Who are the source data producers? Spider: Yale-LILY Lab
CoSQL: Yale-LILY Lab
SparC: Yale-LILY Lab
Custom Dataset: Mudasir Ahmad Mir
Annotations [optional] The SQL queries in this dataset are manually curated and validated by the original dataset creators.
Annotation process Annotations were created as part of the original datasets. For example:
Spider: Annotators were provided with database schemas and asked to write SQL queries for given natural language questions.
CoSQL: Annotators engaged in multi-turn conversations to generate context-dependent SQL queries.
Who are the annotators? The annotators include researchers and contributors from Yale-LILY Lab and other organizations involved in the original datasets.
Personal and Sensitive Information This dataset does not contain personal, sensitive, or private information.
Bias, Risks, and Limitations Recommendations Users should be aware of the following:
The dataset may contain biases inherent in the original datasets, such as domain-specific language or query complexity.
The dataset is not designed for real-time applications without further fine-tuning or optimization.
Citation [optional] If you use this dataset in your research, please cite the original datasets (Spider, CoSQL, SparC, etc.) along with this merged version.
BibTeX:
bibtex Copy @misc{text-to-sql, author = {Mudasir Ahmad Mir}, title = {Text-to-SQL Dataset}, year = {2023}, publisher = {Hugging Face}, howpublished = {\url{https://huggingface.co/datasets/Mudasir692/text-to-sql}} } APA: Mudasir Ahmad Mir. (2023). Merged Text-to-SQL Dataset. Hugging Face. https://huggingface.co/datasets/Mudasir692/text-to-sql
Glossary [optional] Text-to-SQL: The task of converting natural language questions into SQL queries.
Cross-Domain: Refers to datasets that cover multiple domains or topics.
Multi-Turn Interactions: Conversations where SQL queries depend on previous interactions.
More Information [optional] For more information, visit the Hugging Face dataset page: Merged Text-to-SQL Dataset.
Dataset Card Authors [optional] Mudasir Ahmad Mir
Dataset Card Contact For questions or feedback, please contact Mudasir Ahmad Mir.