WebRAG: An Intelligent Website Question Answering System

• A lightweight Retrieval Augmented Generation (RAG) application.
• Enables users to ask natural language questions about the provided website content.
• The system extracts information from webpages, converts them into semantic vector embeddings, retrieves the most relevant context using ChromaDB, and generates accurate, context-aware responses using a locally hosted LLM through Ollama.

Built with LangChain 1.x, ChromaDB, HuggingFace Embeddings, and Ollama.

---

Features

• Retrieval Augmented Generation (RAG)
• Web page ingestion using WebBaseLoader
• Automatic Document chunking
• Semantic search with Chroma Vector Database
• Local inference using Ollama
• HuggingFace Sentence embeddings
• Modular project structure
• No cloud based LLM APIs required

---

Tech Stack

1. Python 3.10+
2. LangChain 1.x
3. ChromaDB
4. Hugging Face Embeddings (BAAI/bge-base-en-v1.5)
5. Ollama
6. Llama 3.2 (1B) (or any compatible Ollama model)

---

Project Structure

WebRAG/
│
├── ingest.py # Builds the vector database
├── chatbot.py # Runs the chatbot
├── chroma_db/ # Persistent Chroma database
├── .env.example
├── .gitignore
├── requirements.txt
└── README.md

---

Installation

1. Clone the repository

git clone <repo-url>
cd WebRAG

2. Create a virtual environment

python3 -m venv rag 
source rag/bin/activate 

3. Install dependencies

pip install -r requirements. txt

4. Download and install Ollama

curl -fsSL https://ollama.com/install.sh | sh

6. Pull an LLM

Example -

ollama pull llama3.2:1b

Environment Variable

Create a .env file in the project root.

Example-

HF_TOKEN=hf_xxxxxxx

Note: The Hugging Face token is only required if you are using hosted Hugging Face services.
Local embedding models typically do not require it once downloaded.

Building the Vector Database

Run the ingestion script once.

~$ python ingest.py

This script:

• Downloads the webpages.
• Splits them into chunks.
• Generates embeddings.
• Stores them in the ChromaDB.

The vector database is saved inside the chroma_db/ directory. You only need to run this again if you change or add source documents.

Running the Chatbot

~$ python chatbot.py

Example:

---

How it Works

---

Future Improvements

• PDF document ingestion
• Multiple website support
• Web Interface or Frontend UI
• Conversation Memory
• Faster responses