Samsung Phone Query & Review System

An system that scrapes real-world Samsung phone specifications, answers natural language queries using LLMs with RAG, generates human-like reviews through a multi-agent system, and exposes all functionality through a modular API.

🚀 Built using Python, BeautifulSoup, PostgreSQL, FastAPI, FAISS, HuggingFace Transformers, and SQLAlchemy.

📌 Table of Contents

• Project Overview
• Features
• Tech Stack
• Project Architecture
• Setup & Usage
• API Endpoints
• Example Queries
• Folder Structure
• Credits

📖 Project Overview

This project was developed as part of a technical assignment to build a query system that enables users to:

• View detailed specifications of Samsung smartphones.
• Ask natural language questions like "Which Samsung phone has the best battery?"
• Read auto-generated product reviews based on real specs.

The system scrapes 10–15 Samsung phones from GSMArena, stores them in a PostgreSQL database, embeds the data using SentenceTransformers, and uses a vector search (FAISS) + LLM approach to answer user queries. A multi-agent system handles review generation, and a clean FastAPI backend powers the final interface.

🎯 Features

✅ Web scraping with BeautifulSoup
✅ PostgreSQL-backed structured data model
✅ RAG-style chatbot using HuggingFace models
✅ FAISS vector search for semantic matching
✅ Multi-agent system for review generation
✅ Modular FastAPI backend with clean routing
✅ Fully testable and extensible architecture

🛠️ Tech Stack

Layer	Tools Used
Web Scraping	BeautifulSoup, requests
Backend Database	PostgreSQL, SQLAlchemy ORM
Embeddings	SentenceTransformer (all-MiniLM-L6)
Vector Index	FAISS
LLM	google/flan-t5-xl (Hugging Face)
Multi-Agent System	Custom-built (DataAgent, ReviewAgent, Coordinator)
API	FastAPI
Testing	Pytest (unit tests per module)

🧠 Project Architecture

📥 Scraper: Collects structured and unstructured Samsung phone data from GSMArena.

🧾 Database: Normalized PostgreSQL schema with Phone and Specification tables.

🧠 RAG Chatbot:

• Embeds phone data with SentenceTransformer
• Stores in FAISS vector index
• Uses vector similarity to retrieve top phones
• Feeds prompt to FLAN-T5-XL to generate answer

🤖 Multi-Agent System:

• DataAgent fetches specs
• ReviewAgent writes a product review
• Coordinator combines both

🌐 API:

• /chatbot/query for open-ended questions
• /review/{phone_name} for detailed specs + review

⚙️ Setup & Usage

Clone the repository

git clone https://github.com/your-username/samsung-phone-query.git
cd samsung-phone-query

Create a virtual environment and install dependencies

python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows
pip install -r requirements.txt

Setup your PostgreSQL database

# Create your DB, then set your credentials in a .env file
cp .env.example .env

Initialize and seed database

python scripts/setup_db.py
python scripts/run_scraper.py
python scripts/seed_data.py

Generate embeddings for chatbot

python chatbot/embeddings.py

Start the FastAPI server

uvicorn api.main:app --reload

Visit http://127.0.0.1:8000/docs to explore the interactive Swagger UI.

🧪 API Endpoints

Method	Route	Description
POST	/chatbot/query	Ask any question about Samsung phones
GET	/review/{phone_name}	Get full phone specs + auto-generated review

💬 Example Queries

Try asking the chatbot:

• "Which Samsung phone has the best battery life?"
• "Compare the Galaxy S23 and S22."
• "What are the camera specs of the Galaxy A55?"

Or use:

curl -X POST http://localhost:8000/chatbot/query \
     -H "Content-Type: application/json" \
     -d '{"question": "Which Samsung phone has the best camera?"}'

📁 Folder Structure

Samsung Phone Query and Review System/
│
├── README.md                      # Project documentation
├── requirements.txt               # Python dependencies
├── .env.example                   # Template for environment variables
├── .gitignore                     # Files/folders to ignore in Git
│
├── config/
│   ├── __init__.py
│   ├── settings.py                # Env configs, constants
│   └── database.py                # SQLAlchemy connection setup
│
├── database/
│   ├── __init__.py
│   ├── models.py                  # SQLAlchemy models
│   └── setup.py                   # DB initialization logic
│
├── scraper/
│   ├── __init__.py
│   ├── gsmarena_scraper.py        # GSMArena scraping logic
│   └── utils.py                   # Scraper helper functions
│
├── chatbot/                       # LLM & RAG system
│   ├── __init__.py
│   ├── embeddings.py              # SentenceTransformer-based embeddings
│   ├── retriever.py               # FAISS retrieval logic
│   ├── prompts.py                 # Prompt templates
│   └── chatbot.py                 # Core chatbot logic (RAG style)
│
├── agents/                        # Multi-agent system
│   ├── __init__.py
│   ├── data_agent.py              # Retrieves specs from DB
│   ├── review_agent.py            # Writes review from specs
│   └── coordinator.py             # Coordinates agents using CrewAI or LangChain
│
├── api/
│   ├── __init__.py
│   ├── main.py                    # FastAPI app entrypoint
│   ├── router.py                  # Combines all sub-routers
│   │
│   ├── chatbot/
│   │   ├── __init__.py
│   │   ├── routes.py              # /query
│   │   ├── schemas.py             # Pydantic models
│   │   └── services.py            # Logic that calls chatbot pipeline
│   │
│   ├── phone_review/
│       ├── __init__.py
│       ├── routes.py              # /review/{phone}
│       ├── schemas.py             # Response schema
│       └── services.py            # Calls coordinator
│
├── tests/
│
├── scripts/
│
└── docs/
    ├── API_GUIDE.md               # Route and usage instructions
    └── SETUP.md                   # Project setup instructions

🙋‍♂️ Credits & Acknowledgements

• GSMArena (data source)
• Hugging Face
• Sentence-Transformers by UKPLab
• FAISS (Facebook AI Similarity Search)

📧 Contact: sadmansakibrafi.hey@gmail.com