Electronic Search Engine

Overview

The Electronic Search Engine is a recommendation system built using Streamlit that helps users find similar electronic products based on their selection. The system leverages a precomputed similarity matrix to provide recommendations.

Features

• User-friendly Streamlit interface.
• Provides top 10 similar products based on selected electronic items.
• Displays product names and images for better visualization.
• Uses a precomputed similarity matrix to ensure quick recommendations.
• Handles image validation before displaying recommendations.

Installation

Prerequisites

Ensure you have the following installed:

• Python 3.x
• Required Python libraries: streamlit, pandas, numpy, nltk, scikit-learn, pickle, PIL, requests

Setup

1. Clone the repository or download the project files.
2. Install the required dependencies:

   pip install streamlit pandas numpy nltk scikit-learn pillow requests

3. Place the data.pkl and similarity.pkl files inside the model directory.

Usage

Run the application using Streamlit:

streamlit run app.py

File Structure

Electronic Search Engine/
│── model/
│   ├── data.pkl
│   ├── similarity.pkl
│── data/
│   ├── All Electronics.csv
│── app.py
│── test.py
│── electronic_search.ipynb
│── README.md

• app.py: Main Streamlit application for product search and recommendations.
• test.py: Additional script for testing image validity before displaying recommendations.
• model/data.pkl: Pickle file containing electronic product data.
• model/similarity.pkl: Precomputed similarity matrix for recommendations.
• model_training.py: Script for processing data, generating similarity matrix, and saving model files.

How It Works

1. Data Preprocessing:

   • Load electronic product data from CSV.
   • Clean and preprocess data (handle missing values, remove duplicates, tokenize text, apply stemming, etc.).
   • Convert text data into a numerical format using Count Vectorization.
   • Compute similarity scores using Cosine Similarity.

2. Recommendation System:

   • User selects a product from the dropdown.
   • The application fetches the top 10 most similar products based on precomputed similarity.
   • The recommended products are displayed with images and names.

Example Output

• User selects: iPhone 12
• Recommended products:
  • iPhone 12 Pro
  • iPhone 11
  • Samsung Galaxy S21
  • etc.

Future Enhancements

• Implement a search bar for quick product lookup.
• Improve similarity calculations using deep learning models.
• Add a feedback mechanism to improve recommendations.

License

This project is open-source under the MIT License.