Skip to content

Latest commit

 

History

History
196 lines (125 loc) · 4.24 KB

README.md

File metadata and controls

196 lines (125 loc) · 4.24 KB

🍕 Local RAG — Restaurant Review Analyzer

Retrieve & Analyze Restaurant Reviews Using Ollama Model + LangChain + ChromaDB

This project is a Local Retrieval-Augmented Generation (RAG) system that answers questions about a restaurant based on realistic customer reviews.

It runs completely offline, powered by:

  • 🧠 Ollama LLMs
  • 🔗 LangChain
  • 🗂 ChromaDB
  • 📊 CSV-based dataset

Ask any question like:

"What do customers think about the pizza crust?" "What are the top complaints?" "Which pizzas are most praised?"

The model answers using actual retrieved reviews.

📁 Project Structure

local-rag/
│── main.py                      # RAG pipeline & chatbot loop  
│── vector.py                    # Vector DB creation + retriever  
│── realistic_restaurant_reviews.csv   # Dataset  
│── requirements.txt             # Dependencies  
│── chrome_langchain_db/         # Auto-created vector DB  
│── venv/                        # Virtual environment

🖼️ Screenshots

📌 1. Dataset Preview (CSV File)

image

📌 2. Project Folder Structure (VS Code)

image

📌 3. CLI RAG Chatbot Running

image image

📌 4. Streamlit RAG Chatbot Running

image image

🚀 Features

🔹 Local RAG Pipeline

Runs without cloud APIs using Ollama.

🔹 Smart Retrieval

Top 5 relevant reviews fetched using vector similarity.

🔹 Persistent Vector DB

Chroma saves embeddings & reloads instantly.

🔹 Uses Real Reviews

Answers directly grounded in CSV data.

🔹 Lightweight & Fast

Perfect for learning + portfolio projects.

🧠 How the RAG System Works

  1. Load CSV (Pandas)
  2. Convert rows → LangChain Documents
  3. Generate embeddings using mxbai-embed-large
  4. Store vectors in ChromaDB
  5. User asks a question
  6. Retriever gets top 5 relevant reviews
  7. Llama 3.2 generates the final answer

🛠️ Installation

1️⃣ Install Ollama

https://ollama.com/download

Models required:

ollama pull llama3.2
ollama pull mxbai-embed-large

2️⃣ Setup Environment

uv venv
source venv/bin/activate    # macOS/Linux
venv\Scripts\activate       # Windows

3️⃣ Install Python Packages

pip install -r requirements.txt

▶️ Run the Project

python main.py

You will see:

====================================
Ask your question (q to quit):
====================================

💬 Example Questions

  • What do customers think about the service?
  • Summarize the positive reviews.
  • What complaints are mentioned most?
  • Give feedback about thin crust pizza.
  • Are there gluten-free options mentioned?

🧰 Tech Stack

Component Purpose
🐍 Python Application logic
🧠 Ollama Local LLM + embeddings
🔗 LangChain RAG chain
🗂 ChromaDB Vector storage
📊 Pandas CSV processing

📄 Dataset

Columns included:

  • Title
  • Date
  • Rating
  • Review

Used for retrieval + summarization.

⭐ Future Enhancements

  • Streamlit UI version
  • Sentiment Analysis
  • Multi-query reasoning (Agentic RAG)
  • Summary Dashboard

❤️ If this helped, give it a ⭐ on GitHub!