ChatDB AI

ChatDB is an intuitive application designed to simplify working with relational data. It allows users to:

• Upload relational data in CSV format to MongoDB (stored as collections) or Amazon RDS.
• Query the uploaded data using human-like queries that are translated into corresponding SQL or MongoDB queries by the backend.
• Uses Gemini 1.5 Flash LLM to generate the queries
• View query results in a tabular format on the dashboard.

---

Features

1. Upload Relational Data

   • Upload a folder containing CSV files.
   • The folder name is treated as the database name.
   • CSV files within the folder are saved as collections (in MongoDB) or tables (in Amazon RDS).

2. Query Execution

   • Write queries in a natural language format (e.g., "Get all customers from California").
   • The backend converts these queries into either:
     • SQL queries for Amazon RDS.
     • MongoDB queries for MongoDB.
   • Results are displayed in a tabular format.

---

Project Setup and Execution

Prerequisites

• Python 3.7 or higher
• MongoDB instance (local or cloud)
• Amazon RDS instance with a MySQL-compatible database

Steps to Run

1. Clone the repository.

2. Navigate to the project directory.

3. Install required Python libraries using:

   pip install -r requirements.txt

4. Set up the .env file in the project directory with the following content:

   # MongoDB Configuration
   MONGODB_URI=<your_mongodb_uri>
   
   # Amazon RDS Configuration
   RDS_HOST=<rds_host>
   RDS_PORT=3306
   RDS_USER=<rds_user>
   RDS_PASSWORD=<rds_password>
   
   API_KEY="GeminiAPIKey"
   

5. Start the application:

   python app.py

   This starts the backend server.

---

Uploading Data

1. Prepare a folder containing all the CSV files.

   • Folder Name: Will be treated as the database name.
   • File Names: Each CSV file name will become a table (in Amazon RDS) or a collection (in MongoDB).

2. Use the dashboard to upload the folder.

   • MongoDB: Each CSV is stored as a collection in a MongoDB database named after the folder.
   • Amazon RDS: Each CSV is stored as a table in a MySQL-compatible database named after the folder.

---

Querying Data

• Use the dashboard to write human-like queries.
• The backend processes the query and executes:
  • SQL queries in Amazon RDS.
  • MongoDB queries in MongoDB.
• Results are displayed in a table format.

---

Technologies Used

• Backend: Python (Flask)
• Database:
  • MongoDB Atlas
  • Amazon RDS (MySQL-compatible)
• Frontend: HTML, CSS, JavaScript
• Query Translation: Regex-based NLP logic for human-like query conversion OR Generated Using Gemini By Few Shot Prompting Method

---

Contributors

• Darshan Rao

  • Set up the backend architecture.
  • Developed and integrated algorithms for handling queries.
  • Implemented API calls for communication between the frontend and backend.
  • Configured and managed Amazon RDS and MongoDB connections.
  • Added the Query Generation using Gemini Using Few-Shot Prompting

• Shardul Datar

  • Designed and developed the interactive user interface using HTML, CSS, and JavaScript.
  • Implemented features for uploading entire databases and typing in human-like queries.

• Ekaterina Shtyrkova

  • Built the query transformation logic for converting natural-language-like queries into SQL and MongoDB queries.
  • Developed regex templates to map user queries to corresponding database query patterns.

---

Folder Structure

├── app.py                  # Main backend script
├── requirements.txt        # Python dependencies
├── .env.example            # Example environment file
├── uploads/                # Temporary storage for uploaded data
└── README.md               # Project documentation

---

Future Enhancements

• Add support for more database systems (e.g., PostgreSQL, DynamoDB).
• Improve NLP capabilities for better query understanding.
• Implement query history and analytics features.

---

Feel free to contribute by submitting issues or pull requests!