Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Recurso digital |
| Language: | English |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.18315828 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901351200456704 |
|---|---|
| author | Vishesh Jain Aryan Sharma Shubham Saini Gaurav Yadav |
| author_facet | Vishesh Jain Aryan Sharma Shubham Saini Gaurav Yadav |
| contents | <p>This work presents an offline Text-to-SQL system that enables users to query relational databases using natural language without requiring cloud-based large language models. The system combines a locally deployed Qwen2.5 7B model with a modular backend built using FastAPI, a Streamlit-based interface for interaction, and PostgreSQL for secure query execution. The research emphasizes privacy, data security, and operational independence by eliminating external API calls and internet dependencies.</p> <p>The proposed pipeline integrates schema extraction, prompt engineering, validation, and SQL execution to provide accurate syntactic and semantic mapping between user queries and SQL. Experimental evaluation shows high accuracy on simple and moderately complex SQL tasks, demonstrating feasibility for real-world use cases in regulated or sensitive environments.</p> <p>The paper also analyzes existing research gaps in Natural Language Interfaces to Databases (NLIDBs), compares offline LLM-driven inference with cloud solutions, and outlines future enhancements including multilingual support, conversational context, cross-database interoperability, and visualization.</p> <p>This contribution is intended for researchers, developers, and practitioners working on NLIDBs, enterprise automation, offline LLM deployment, and privacy-preserving AI systems. It highlights a practical direction for building secure, self-contained natural language database querying tools suitable for enterprise, academic, and industrial settings.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18315828 |
| institution | Zenodo |
| language | eng |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | AI-Powered Natural Language to SQL Transformation: A Secure Offline Architecture for Intelligent Database Querying Vishesh Jain Aryan Sharma Shubham Saini Gaurav Yadav <p>This work presents an offline Text-to-SQL system that enables users to query relational databases using natural language without requiring cloud-based large language models. The system combines a locally deployed Qwen2.5 7B model with a modular backend built using FastAPI, a Streamlit-based interface for interaction, and PostgreSQL for secure query execution. The research emphasizes privacy, data security, and operational independence by eliminating external API calls and internet dependencies.</p> <p>The proposed pipeline integrates schema extraction, prompt engineering, validation, and SQL execution to provide accurate syntactic and semantic mapping between user queries and SQL. Experimental evaluation shows high accuracy on simple and moderately complex SQL tasks, demonstrating feasibility for real-world use cases in regulated or sensitive environments.</p> <p>The paper also analyzes existing research gaps in Natural Language Interfaces to Databases (NLIDBs), compares offline LLM-driven inference with cloud solutions, and outlines future enhancements including multilingual support, conversational context, cross-database interoperability, and visualization.</p> <p>This contribution is intended for researchers, developers, and practitioners working on NLIDBs, enterprise automation, offline LLM deployment, and privacy-preserving AI systems. It highlights a practical direction for building secure, self-contained natural language database querying tools suitable for enterprise, academic, and industrial settings.</p> |
| title | AI-Powered Natural Language to SQL Transformation: A Secure Offline Architecture for Intelligent Database Querying |
| url | https://doi.org/10.5281/zenodo.18315828 |