Mohawk College Conversational Assistant 🏫

Welcome to the Mohawk College Conversational Assistant! This is a FastAPI-based backend service designed to handle user queries about Mohawk College programs, student services, and more. It uses LangChain for Retrieval-Augmented Generation (RAG), OpenAI for text processing, and MongoDB for storing conversation history.

✨ Key Features

• Retrieval-Augmented Generation (RAG) using local documents for more accurate, context-specific answers.
• OpenAI GPT integration for query refinement and fallback responses.
• MongoDB Storage for logging user queries and responses.
• WebSocket Support for real-time interactive conversations.
• External Crawling to fetch content from Mohawk College or MSA websites when queries pertain to specific keywords.

📂 Table of Contents

1. Prerequisites
2. Project Setup
3. Environment Variables
4. Run the Application
5. API Endpoints
6. WebSocket Usage
7. Database Integration
8. Directory Structure

---

🛠 Prerequisites

• Python 3.9+ (Recommended)
• pip or conda for installing Python packages
• MongoDB instance (local or remote)
• An OpenAI API key (for GPT-3.5-turbo or GPT-4 calls)

---

🏗 Project Setup

1. Clone the repository:

   git clone https://github.com/temi-t8/backend-server.git

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

python -m venv venv
source venv/bin/activate   # On Linux/Mac
venv\Scripts\activate      # On Windows

3. Install dependencies

pip install -r requirements.txt

4. 🔐 Environment Variables

export OPENAI_API_KEY=sk-proj
export MONGODB_URI="mongodb+srv://TEMI:robot@"

NOTE: You can create your own MongoDB Cluster and add MongoDB URI but make sure to keep double inverted comma to avoid error

🚀 Run the Application

After installing all dependencies and setting your environment variables, you can start the server:

uvicorn app:app --host 0.0.0.0 --port 8000 --reload

The application will be available at: http://localhost:8000

📡 API Endpoints

1. Health Check

• Endpoint: GET /health
• Description: Returns the current status of the service and MongoDB connection.
• Response Example:

  {
  "status": "running",
  "version": "1.0.0",
  "db_status": "healthy"
  }

2. Answer Query

• Endpoint: POST /answer

• Request Body:

  {
  "query": "What programs does Mohawk College offer related to Software?"
  }

• Response Example:

  {
  "answer": "Mohawk College offers the Computer Systems Technician – Software Development program..."
  }

• Description:

  • Refines the user query for spelling or partial program names.
  • Checks if the query relates to Mohawk programs, MSA, admissions, etc.
  • Crawls official Mohawk/related sites for matching content or uses Local RAG from the data folder to generate an answer.
  • If no context found, it uses a fallback with OpenAI GPT to provide a generic or playful response.

⚡ WebSocket Usage

• Endpoint: GET /ws
• Protocol: WebSocket

This endpoint allows real-time conversational interaction:

1. Connect to the WebSocket at ws://localhost:8000/ws.
2. Send a message (the user query).
3. Receive a text response.

Example (using websocat or similar CLI tool):

websocat ws://localhost:8000/ws
> Hello, what jokes do you know?
< Here's a fun one...

The server will:

• Refine your query.
• Check if it’s a fun/joke request or a Mohawk-related query.
• Respond with either RAG-based or fallback content.
• Store the conversation in MongoDB.

NOTE: if this server is being deployed anywhere it should use secure websocket such as wss://domain/ws/

🌐 Database Integration

This service uses MongoDB to record each user query and the corresponding answer.

• Collection Name: ConversationTracker
• Stored Fields:
  • query: The original user query.
  • answer: The assistant’s answer.
  • timestamp: DateTime of the stored record.

🗂 Directory Structure

.
├── data/
│   ├── faqs.txt
│   └── ...
├── main.py
├── requirements.txt
├── .gitignore
└── README.md

If you have any questions, reach out via email me at nevil-dineshkumar.patel@mohawkcollege.ca