Recommendation System

Overview

This project is a Recommendation System that allows users to:

1. Insert product data into MongoDB and Qdrant vector database.
2. Search for similar products using vector-based similarity search.
3. Filter results based on attributes like color and price.
4. Analyze Transactions to suggest related products based on purchasing behavior.
5. Search Online for related items from Amazon using Tavily API.

---

Features

• MongoDB Integration: Stores product and transaction data.
• Vector Search with Qdrant: Embeds product descriptions and enables similarity-based retrieval.
• User Transactions Analysis: Identifies frequently bought-together items.
• Tavily Web Search: Fetches related product listings from Amazon.
• FastAPI Backend: Provides endpoints for inserting data, searching, and recommendations.

---

Tech Stack

• FastAPI - Backend framework
• MongoDB - NoSQL database for storing products & transactions
• Qdrant - Vector database for similarity search
• Tavily API - Online search integration
• Python 3.12 - Programming language

---

Installation

1️⃣ Clone the Repository

git clone https://github.com/yourusername/recommendation-system.git
cd recommendation-system

2️⃣ Set Up Virtual Environment

python -m venv venv
source venv/bin/activate  # On Windows, use: venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

---

Configuration

1️⃣ Set Up Environment Variables

Create a .env file in the root directory:

QDRANT_URI="https://"
QDRANT_API_KEY="euJhbGc**************"
MONGO_URI="mongodb://"
COHERE_API_KEY="vrx0*****************"
VECTOR_DB_PORT=6***
VECTOR_DB_URI="http:/"
TAVILYAPI_KEY="tvly-dev-*************"
MONGO_DB_NAME=""

2️⃣ Start MongoDB & Qdrant

Make sure MongoDB and Qdrant are running locally or on your cloud provider.

3️⃣ Run the Application

uvicorn app.main:app --reload

---

API Endpoints

1️⃣ Insert Item

POST /insert_item

Request Body (JSON):

{
  "name": "Yale Forklift",
  "description": "رافعات شوكية لنقل المواد الثقيلة",
  "material": "Metal",
  "color": "Yellow",
  "price": 250000.0,
  "related_items": ["Pallet Jack", "Digital Scale"],
  "category": "Heavy Machinery"
}

2️⃣ Search for Similar Items

GET /search?query=Forklift&filter_color=Yellow&filter_price_max=300000

3️⃣ Get Related Items from Transactions

GET /related_items/{item_id}

4️⃣ Search Amazon for Related Products

GET /web_search?query=Forklift site:amazon.com

---

Future Enhancements:

• 🔹 Enhance Recommendation Algorithms: Implement collaborative filtering techniques to offer more personalized product suggestions based on user behavior.

• 🔹 ML-Powered Product Prediction: Develop a machine learning model to predict and recommend products to users based on their preferences and historical data.

• 🔹 Front-End Integration: Build and integrate a user-friendly interface to display recommendations and interact with the system seamlessly.

• 🔹 Support Multiple Languages: Add multi-language support to broaden accessibility and make the platform user-friendly across different regions.