Natural Language to SQL Framework: A Streamlit-based Interface for Non-Technical Users

The objective of this paper is to develop an AIpowered system for translating natural language into SQL queries, intended to streamline database interactions for nontechnical users. The proposed system generates executable SQL statements from user inputs by leveraging the Gemini API, which naturally facilitates retrieval augmented generation. A Streamlit-based interface enables users to upload structured Excel datasets and execute Create, Read, Update and Delete (CRUD) operations using conversational commands. The process encompasses schema parsing, sample data extraction, and prompt engineering to ease query generation. When analyzed using a student information dataset, the system attained a query generation accuracy of 92% and an execution success rate of 90 %. Use case scenarios such as data insertion, filtering, updating, and deletion illustrate the effectiveness of the model in understanding the user intent. Despite its dependable functioning, issues such as response latency and sporadic misreading of natural language inquiries persist. This study illustrates the practical capabilities of large language models, like Gemini, in connecting natural language with structured data access, thus improving database interactions.