RoostAI: A University-Centered Chatbot

Overview

RoostAI is an intelligent chatbot specifically designed to provide information about the University of South Carolina (USC). It uses advanced RAG (Retrieval-Augmented Generation) techniques to deliver accurate, context-aware responses to university-related queries. Check out a presentation of the application's functionality

Project Structure

roostai/            
├── back_end/      # Core chatbot implementation
├── front_end/     # User interface components
├── web_scraping/  # Data collection utilities
├── scripts/       # Utility scripts for setup and maintenance
└── data/          # Vector database and related data

eval/              # Evaluation scripts and tools

Setup and Installation

1. Prerequisites

   • Python 3.9+
   • Poetry for dependency management
   • HuggingFace API key for LLM access
   • Optional: Might need OpenAI, Anthropic, and Google API keys if considering comprehensive evaluations

2. Installation

poetry install

3. Environment Setup Add the following environment variables to your .env file:

# HuggingFace API Key
HF_API_KEY=your_huggingface_api_key

# Optional: OpenAI, Anthropic, and Google API Keys
OPENAI_API_KEY=your_openai_api_key
ANTHROPIC_API_KEY=your_anthropic_api_key
GOOGLE_API_KEY=your_google_api_key

Usage Commands

All commands should be run from the project root directory.

Data Collection and Processing

# Web Scraping Pipeline
poetry run python roostai/web_scraping/web_scraper.py
poetry run python roostai/web_scraping/collect_html.py
poetry run python roostai/web_scraping/extract.py
poetry run python roostai/web_scraping/chunker.py

# Database Creation
poetry run python roostai/scripts/data_ingestion.py

Running the Chatbot

# Interactive Mode
poetry run python roostai/back_end/main.py

# With Output Logging
poetry run python roostai/back_end/main.py 2>&1 | tee roostai/back_end/dry-run.out

Evaluation Tools

# FAQ Response Generation
poetry run python eval/get_llm_responses.py

# RAG Response Evaluation
poetry run python eval/ragas_evaluation/get_rag_response.py

Project Components

Back-End

• RAG-based retrieval system
• Vector database management
• Query processing and response generation
• Quality assessment and reranking

Front-End

• User interface for chatbot interaction using Streamlit
• Survey functionality for user feedback
• Response visualization

Data Processing

• Web scraping utilities
• Text chunking and preprocessing
• Data ingestion pipeline

License

This project is licensed under the BSD-2 Clause License. For more information, see the LICENSE file.