University of New Haven AI FAQ Chatbot

Overview

The University of New Haven FAQ Chatbot is a sophisticated tool designed to provide quick and accurate responses to student inquiries. Utilizing LangChain, a large language model, and Pinecone, a vector database, this chatbot stands out in its efficiency and reliability. Check out the project here.

App Architecture

The architecture of the application ensures seamless integration of various technologies for an optimal user experience.

Project Structure

The project is organized into several key components:

• main flask app.py: The core Flask server and web application, managing user interactions and the chat interface.
• loaddata.py: Responsible for scraping, processing, and populating the Pinecone database with necessary data.
• UI Components: Includes static files and templates for the chatbot's user interface.

Prerequisites

To get started, you'll need:

• Python 3.8 or higher
• Flask
• MongoDB account and cluster
• Pinecone account
• OpenAI API key
• A .env file with necessary environment variables

Setup Instructions

1. Environment Variables

Set up your .env file with:

PINECONE_API_KEY=your_pinecone_api_key
PINECONE_ENVIRONMENT=your_pinecone_environment
PINECONE_INDEX=your_pinecone_index_name

2. MongoDB Configuration

• Create a MongoDB database and collection for storing conversations.
• Update the MongoDB URI in main flask app.py.

3. Pinecone Configuration

• Set up a Pinecone account and create an index.
• Update your .env file with Pinecone details.

4. Install Dependencies

Run pip install –r requirements.txt to install all necessary Python packages.

Running the Application

Data Loading

Optional: The data may already be loaded.

• Execute loaddata.py to scrape, process, and load FAQs into Pinecone.

Starting the Flask App

• Run python app.py to start the Flask server.
• Alternatively, use Docker:
  • Build the image: docker build -t unh-faq-bot .
  • Run the container: docker run -p 5000:5000 unh-faq-bot

Accessing the Web Interface

• Visit http://localhost:5000/ in your browser to use the chatbot.

Usage

• Users can ask questions via the chat interface.
• The bot provides answers with sources, when available.

Managing Data

• Conversation data is stored in MongoDB.
• Utilize MongoDB's tools for data management and analysis.

Troubleshooting

If issues occur, check:

• Environment variables in .env.
• MongoDB URI and database accessibility.
• Pinecone index configuration and permissions.
• Validity of Pinecone and OpenAI API keys.

Conclusion

This provides a detailed guide for setting up and running the University of New Haven FAQ Chatbot. For support or queries, contact Krishna Vamsi Nadh.