Talk to Database App – Text2Query

Talk to Database App – Text2Query
View Report

Project Overview

Talk to Database App (Text2Query) is a web-based tool designed to convert natural language statements into MySQL queries. This project aims to assist non-technical users, such as analysts and business professionals, in retrieving database information without SQL expertise.

My Role

  • Product Management: Defined scope, roadmap, and user requirements.
  • Market Research & Stakeholder Analysis: Identified target users and business use cases.
  • Technology Evaluation: Analyzed and selected Flask, React.js, and MySQL for development.
  • Development & Testing: Assisted in rule-based model creation and UI development.
  • User Feedback & Iteration: Gathered insights from stakeholders to refine functionality.

Key Outcomes

1. Problem Statement & Market Need

  • Current Challenge: Business users depend on data analysts for SQL queries, causing inefficiencies.
  • Solution: Automate query generation by converting English text into MySQL queries.
  • Target Users: Data analysts, financial analysts, customer support teams, and business executives.

2. Features

  • Natural Language Input → Users can type requests in plain English.
  • Rule-Based Query Generation → The system maps input text to MySQL queries.
  • Remote Query Execution → Fetches data directly from a live database.
  • Data Preview & Export → Allows downloading query results in various formats.

3. Technology Stack

Component Technology
Frontend React.js, JSX
Backend Flask, Python
Database MySQL, SQLAlchemy
Libraries ReGex, CSV, IO

4. Evolution Over Time

Phase 1: Deep Learning Approach (Abandoned)

  • Challenge: Training an NLP model required extensive data, resources, and expertise.
  • Decision: Shifted to a pre-trained NLP model for better efficiency.

Phase 2: Using Pre-Trained Models

  • Advantage: Improved query conversion accuracy.
  • Challenge: Handling complex queries and schema mismatches.
  • Decision: Pivoted to Rule-Based Parsing for better control & consistency.

Phase 3: Rule-Based System (Final Approach)

  • Outcome: A deterministic system with predefined SQL query structures.
  • Challenges Solved:
    • Ambiguous user inputs.
    • Handling structured MySQL data.
    • Ensuring query accuracy & error handling.

5. Stakeholders & User Personas

User Role Needs
Soumya Data Analyst Wants to reduce time spent writing queries manually.
Virupakshi Loan Analyst Needs quick database insights using English text.
Sarika, Radha & Swaroop Field Operators Require data for premium recovery follow-ups.

6. User Journey

  • Step 1: User inputs a question in natural language.
  • Step 2: System processes text and maps it to an SQL query.
  • Step 3: The query executes on the MySQL database.
  • Step 4: Users can preview or download the retrieved data.

7. Challenges & Learnings

Challenges Faced

  • Handling ambiguous queries.
  • Building scalable SQL parsing rules.
  • Ensuring non-technical users could understand results.

Learnings

  • Rule-based systems offer better control over structured queries.
  • Stakeholder feedback is critical for refining usability.
  • Edge case handling is essential for real-world applications.

8. Contribution Breakdown

Team Member Role Contribution
Aditya Singh Product Manager System design, Documentation, UI design & implementation.
Aditya Sangana Frontend Engineer React.js development, UI implementation.
M. L. Raghurama Datta Backend Engineer Rule-based model, API development.
Sandeep Nagalli Backend Engineer MySQL setup, query processing.

9. Demo & Source Code

This project demonstrates how natural language processing can bridge the gap between non-technical users and SQL databases, improving data accessibility and efficiency.