Book Recommender System

A intelligent book recommendation web application built with Flask and machine learning. This system suggests similar books based on user input and displays the top 50 most popular books from a comprehensive dataset.

📚 Overview

The Book Recommender System is designed to help users discover new books they might enjoy based on similarity analysis. It features two main sections:

1. Home Page: Displays the top 50 most rated and voted books with book covers, titles, authors, vote counts, and ratings
2. Recommendation Engine: Allows users to search for a book and receive personalized recommendations based on cosine similarity scores

🎯 Features

• Popular Books Showcase: Browse the top 50 most popular books with detailed information including:

  • Book cover images
  • Book titles
  • Author names
  • Number of votes/ratings
  • Average rating scores

• Intelligent Recommendations: Search for any book and receive 4 personalized book recommendations based on collaborative filtering using cosine similarity

• Interactive Web Interface: Clean, user-friendly interface built with Bootstrap with a dark theme and green navigation bar

• Fast Recommendations: Pre-computed similarity scores for instant book recommendations without model retraining

🏗️ Project Structure

book_recommender_system/
├── app.py                 # Flask application with routes
├── book_recommender.ipynb # Jupyter notebook with model development
├── requirements.txt       # Python dependencies
├── pyproject.toml        # Project configuration
├── Procfile              # Heroku deployment configuration
├── README.md             # This file
├── model/                # Pre-trained model files
│   ├── popular.pkl       # Top 50 popular books dataset
│   ├── pt.pkl            # Pivot table for recommendations
│   ├── books.pkl         # Complete books dataset
│   └── similarity_scores.pkl  # Pre-computed similarity matrix
└── templates/            # HTML templates
    ├── index.html        # Home page with top books
    └── recommend.html    # Recommendation page

🛠️ Technology Stack

• Backend: Flask
• Data Processing: Pandas, NumPy
• Machine Learning: Cosine Similarity (Collaborative Filtering)
• Frontend: Bootstrap 3.7
• Data Serialization: Pickle

📋 Requirements

Flask
NumPy
Pandas

🚀 Getting Started

Prerequisites

• Python 3.7 or higher
• pip (Python package manager)

Installation

1. Clone or download the project:

cd book_recommender_system

2. Install dependencies:

pip install -r requirements.txt

3. Run the Flask application:

python app.py

4. Open your browser and navigate to:

http://127.0.0.1:5000/

💻 Usage

Home Page

Navigate to the home page to view the Top 50 Books dashboard:

The homepage displays:

• Book cover images in a card layout
• Book titles and authors
• Number of votes received
• Average rating scores
• Easy navigation to the recommendation engine

Recommendation Page

Click on "Recommend" in the navigation bar to access the recommendation engine:

How to use:

1. Enter a book title in the search box
2. Click the "Submit" button
3. The system displays 4 recommended books similar to your search
4. Recommendations are based on book similarity and user ratings

🤖 How It Works

Data Processing Pipeline

1. Data Collection: The system uses a dataset containing books with their titles, authors, and user ratings
2. Popular Books Selection: Identifies the top 50 most voted and rated books for homepage display
3. Pivot Table Creation: Creates a user-book rating matrix for similarity computation
4. Similarity Calculation: Computes cosine similarity scores between all books
5. Recommendations: Returns the 4 most similar books (excluding the query book itself)

Algorithm

The recommendation engine uses Cosine Similarity with Collaborative Filtering:

• Cosine Similarity: Measures the angle between rating vectors, ignoring magnitude
• Collaborative Filtering: Based on the assumption that users who rated books similarly will enjoy the same books
• Pre-computed Scores: Similarity matrices are pre-calculated for instant recommendations

📊 Model Files

The model/ directory contains pre-trained pickle files:

File	Purpose
popular.pkl	DataFrame of top 50 popular books
pt.pkl	Pivot table (books × users) for similarity computation
books.pkl	Complete books dataset with metadata
similarity_scores.pkl	Pre-computed cosine similarity matrix

📝 API Routes

GET /

• Description: Home page showing top 50 books
• Response: HTML page with book gallery

GET /recommend

• Description: Recommendation search interface
• Response: HTML form for book search

POST /recommend_books

• Description: Get recommendations for a book
• Parameters:
  • user_input (string): Book title to search
• Response: HTML page with 4 recommended books

🎨 Frontend Features

• Responsive Design: Works on desktop and mobile devices (Bootstrap grid)
• Dark Theme: Black background with white text for better readability
• Color Scheme: Green navigation bar (#00a65a) and orange submit button
• Card Layout: Books displayed in responsive card components
• Image Integration: Book cover images loaded from URLs

📈 Performance

• Instant Recommendations: Pre-computed similarity scores eliminate model inference delays
• Scalable Architecture: Pickle serialization allows quick loading of large datasets
• Lightweight: No heavy ML libraries required at runtime
• Efficient Search: Direct index lookup for user queries
• Local-First: Optimized for local development and usage

🔄 Data Flow

User Input (Book Title)
    ↓
Index Lookup in Pivot Table
    ↓
Retrieve Pre-computed Similarity Scores
    ↓
Sort by Similarity (descending)
    ↓
Select Top 4 Recommendations
    ↓
Fetch Book Details (title, author, image)
    ↓
Display to User

🐛 Troubleshooting

"Book not found" error

• Ensure the book title exists in the dataset
• Try searching with partial titles
• Check case sensitivity (though the system should handle this)

Missing model files

• Ensure all .pkl files are present in the model/ directory
• Run the Jupyter notebook to regenerate model files if needed

Port already in use

• Change the port in app.py:

app.run(debug=True, port=5001)

📚 Dataset Information

The project uses a comprehensive book dataset containing:

• Thousands of unique books
• User ratings and votes
• Book metadata (title, author, ISBN)
• Book cover images

🔮 Future Enhancements

• User authentication and personalized recommendations
• Advanced filtering (genre, publication year, price range)
• User ratings and review system
• Content-based filtering using book descriptions
• Hybrid recommendation system combining multiple approaches
• Search autocomplete and suggestions
• Mobile app version
• Integration with library APIs

📄 License

This project is open source and available for educational and commercial use.

👤 Author

Created as a book recommendation system for discovering similar books based on user preferences.

📞 Support

For issues or feature requests, please create an issue in the project repository.

---

Last Updated: January 2026