Information Retrieval Chatbot with FastAPI and LangChain

This project is an information retrieval chatbot built using FastAPI and LangChain. It utilizes the OpenAI API for answering user queries based on a set of documents (PDF files). The chatbot can return context-aware answers, extract specific information from documents, and engage in friendly, human-like conversations with users.

Features

• Document-based Q&A: The chatbot retrieves relevant information from uploaded documents (PDF format) and answers user queries.
• Social Interaction: The chatbot is able to respond to common social greetings like "Hi", "Hello", "How are you?", and more in a human-like and friendly manner.
• FastAPI-powered API: The project exposes an API endpoint for querying, which is built using FastAPI, a fast and modern web framework for Python.
• Customizable Instructions: The system allows setting an initial prompt for context that helps the AI assistant answer more effectively and in a natural tone.

Requirements

• Python: 3.8 or higher
• Dependencies:
  • FastAPI
  • Uvicorn
  • LangChain
  • LangChain-OpenAI
  • FAISS (for efficient vector-based search)
  • python-dotenv

Setup Instructions

1. Clone this repository to your local machine:

   git clone https://github.com/levietkhanh189/timo-info-bot-prototype
   cd timo-info-bot-prototype

2. Create a virtual environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate  # For Windows, use `venv\Scripts\activate`

3. Install the required dependencies:

   pip install -r requirements.txt

4. Set your OpenAI API key as an environment variable:

   export OPENAI_API_KEY="your_openai_api_key"  # On Windows use `set` instead of `export`

5. Ensure that your PDF documents are placed in a data folder inside the project directory.

6. Start the FastAPI server:

   uvicorn main:app --reload

7. The API will be available at http://127.0.0.1:8000. You can check the health status of the API at http://127.0.0.1:8000/healthcheck.

API Endpoints

• POST /ask: Submit a query to the chatbot.

  • Request body:

    {
      "query": "Your question here"
    }

  • Response:

    {
      "question": "Your question here",
      "answer": "Chatbot's answer",
      "source_documents": [List of documents referenced]
    }

• GET /healthcheck: Check the health status of the server.

  • Response:

    {
      "status": "ok"
    }

Example Query

Example request:

curl -X POST "http://127.0.0.1:8000/ask" -H "Content-Type: application/json" -d '{"query": "What is the name of the author in the document?"}'

Notes

• This project is designed to retrieve information from PDF files placed in the data folder. You can easily extend it by adding more PDF documents or adapting the document loader.
• The chatbot uses OpenAI’s language model to process the queries and provide natural-sounding responses.
• Make sure to keep your OpenAI API key secure, and do not expose it in public repositories.