---

PRESERVING DATABASE SCHEMA PRIVACY WHILE GENERATING SQL QUERIES USING GENERATIVE AI

Abstract In Generative AI (GenAI) models have proven effective in translating natural language queries into SQL. However, directly exposing database schema details, including table and column names, to GenAI poses significant security and privacy concerns. This paper presents an approach to preserve database schema integrity by employing aliasing techniques. Aliased representations of schema elements are enriched with semantic metadata, ensuring GenAI can generate accurate SQL queries without direct access to the original schema. The proposed method bridges the gap between privacy and functionality, providing a robust framework for secure and efficient SQL generation. Authors Jitesh Prasad Khatick1, Soumitra Kumar Mandal2 Maulana Abul Kalam Azad University of Technology, India1, National Institute of Technical Teachers’ Training and Research, India2 Keywords NLIDB (Natural Language Interface to Database), SQL (Structured Query Language), StanfordCoreNLP, Large Language Model (LLM), Generative AI (GenAI) Yearly Full Views January February March April May June July August September October November December 8 18 2 0 0 0 0 0 0 0 0 0

Published By : ICTACT Published In : ICTACT Journal on Soft Computing ( Volume: 15 , Issue: 3 , Pages: 3662 - 3668 ) Date of Publication : January 2025 DOI : 10.21917/ijsc.2025.0508 Page Views : 173 Full Text Views : 28