Syrian-Heritage-Conversatinal-RAG-Bot

Overview

This project is an AI-powered conversational chatbot designed to answer queries about Syrian heritage. It leverages state-of-the-art natural language processing techniques, retrieval-augmented generation (RAG), and real-time web search. Built with modular architecture, it processes user queries intelligently, providing context-aware responses. The chatbot is integrated with FastAPI for backend services and Streamlit for a user interface.

Project Structure

AI_AGENT_FINAL_VERSION/
│── API/
│   │── __init__.py
│   │── app.py
│   │── schemas.py
│── config/
│   │── __init__.py
│   │── constants.py
│   │── prompts.py
│   │── settings.py
│── core/
│   │── __init__.py
│   │── graph.py
│   │── routers.py
│── data_scripts/
│   │── dataset_generation/
│   │── vector_db_generation/
│   │── __init__.py
│── models/
│   │── __init__.py
│   │── schemas.py
│── notebooks/
│   │── train_embeddings_intent.ipynb
│── services/
│   │── __init__.py
│   │── generator.py
│   │── grader.py
│   │── intent_classifier.py
│   │── retriever.py
│   │── search.py
│   │── summarizer.py
│── user_interface/
│   │── __init__.py
│   │── user_interface.py
│── utils/
│   │── __init__.py
│   │── utils.py
│── docs/
│   │── report_doc.pdf
│── requirements.txt
│── readme.md
│── LICENCE 

Prerequisites

Before running the system, make sure to install all the required dependencies. You can use pip to install them:

1. Install dependencies:

   pip install -r requirements.txt
   
   

2. Create a .env file in the AI_AGENT_FINAL_VERSION directory with the following API keys:

   export GROQ_API_KEY="your_api_key"
   export TAVILY_API_KEY="your_api_key"
   
   

Running the Application

1. Start the FastAPI Server

In one terminal window, navigate to the AI_AGENT_FINAL_VERSION directory and run the FastAPI server with the following command:

python -m uvicorn AI_Agent_final_version.API.app:app --host 0.0.0.0 --port 8000 --reload

This will start the FastAPI server, allowing backend requests to be handled.

2. Launch the User Interface

In another terminal window, navigate to the AI_AGENT_FINAL_VERSION/user_interface directory and run the Streamlit user interface with:

python -m streamlit run AI_Agent_final_version/user_interface/user_interface.py

This will launch the user interface where users can interact with the chatbot.

System Architecture

The system is built around a directed acyclic graph managed by LangGraph. It consists of the following main components:

• Intent Classifier: Determines the nature of the user query (whether it needs to be answered directly or requires further retrieval).
• Retriever (RAG): Retrieves relevant documents from a pre-built FAISS vector database using embeddings.
• Grader: Evaluates the relevance of retrieved documents to the user's query.
• Web Search: Gathers real-time information via the Tavily and DuckDuckGo APIs.
• Generator: Forms the final response based on retrieved documents, web results, and conversation history.
• Summarizer: Updates the conversation memory to maintain context between interactions.

How It Works

1. User Input: The user submits a query to the chatbot through the Streamlit interface.
2. Intent Classification: The query is processed by the intent classifier, which determines if the query requires a retrieval-augmented generation process.
3. Document Retrieval: If necessary, the retriever fetches relevant documents from the vector database.
4. Grading: The grader evaluates the relevance of the retrieved documents, and if they are insufficient, it triggers a web search.
5. Web Search: External search results are fetched in real-time to enhance the response.
6. Response Generation: The final response is generated by combining the query, relevant documents, and any real-time web data.
7. Summarization: The conversation summary is updated to maintain context for future interactions.

Environment Setup

1. .env File: Create the .env file in the AI_AGENT_FINAL_VERSION directory with the required API keys:

   export GROQ_API_KEY="your_api_key"
   export TAVILY_API_KEY="your_api_key"
   
   

2. Dependencies: Install the dependencies listed in requirements.txt using pip.

Requirements

• Python 3.7+
• FastAPI
• Uvicorn
• Streamlit
• Hugging Face
• FAISS
• Tavily API (for real-time search)

Conclusion

This project provides an AI-powered chatbot that educates and engages users with information about Syrian heritage. By utilizing modern NLP techniques, retrieval-augmented generation, and real-time web searches, the system offers accurate, context-aware, and dynamic responses.