Back to Home
AtjBot

AtjBot

GitHub

This project is a multilingual, RAG-powered chatbot application built with a React frontend and a Flask backend. It allows users to chat with a bot that retrieves information from documents and generates responses intelligently. It supports different document types and languages, and stores embeddings using the Chroma vector database.

Key Features

  • Interactive chatbot that responds to user queries
  • Language selector enabling multilingual support
  • Predefined suggestions to guide the conversation
  • RAG (Retrieval-Augmented Generation) pipeline for accurate and contextual answers
  • Extracts and displays reference links used in responses
  • Enforces rate limiting to prevent abuse of the API

Tech Stack

Frontend

  • Built with React for UI
  • Supports modern JavaScript workflows
  • Runs on a local development server

Backend

  • Built with Flask to handle API requests
  • Integrates with LangChain for the RAG pipeline
  • Uses NLTK for NLP tasks
  • Communicates with Google Generative AI for response generation

Storage and Processing

  • Vector storage with Chroma for document embeddings
  • Supports ingestion of PDFs, CSVs, and plain text files

Deployment

  • Docker-compatible architecture for containerized deployment

Installation Summary

  • Requires Node.js and npm for the frontend
  • Requires Python 3.7+ and virtual environment setup for the backend
  • Uses pip to install backend dependencies
  • Requires a Google API key for access to generative AI

Usage Instructions

  1. Launch frontend and backend locally
  2. Access the application via browser
  3. Select a language from the dropdown
  4. Enter a query in the input field and submit
  5. View chatbot's response along with relevant context and links

API Overview

Main Endpoints

  • GET / serves the main frontend
  • POST / predict accepts a user message and returns a generated response

Rate Limiting

  • The / predict endpoint allows up to 10 requests per minute per user

Example Use

  • A user inputs a message with an optional language tag
  • The server processes the query using document context and returns an appropriate response

Contribution Guide

  • Contributions are encouraged
  • Submit issues or pull requests for improvements or bug fixes

License

  • Licensed under the MIT License

Acknowledgments

  • LangChain for enabling document-based retrieval and embedding
  • NLTK for language processing tools
  • React and Flask for frontend and backend frameworks