📚 Unbound – Book Recommendation System

Unbound is a content-based & collaborative filtering book recommendation system built with Python (Flask) for the backend and a lightweight HTML/CSS/JavaScript frontend.
It suggests books based on user-selected titles using K-Nearest Neighbors (KNN) on a sparse ratings matrix.

---

🚀 Features

• Real-time book search with autocomplete suggestions
• Personalized recommendations using collaborative filtering
• Data preprocessing & filtering for cleaner results
• Responsive frontend for smooth user experience
• Architecture page with dataset & filtering insights

---

📊 Dataset

We use the Book-Crossing dataset which contains:

• Books metadata (BX-Books.csv)
• User information (BX-Users.csv)
• Ratings data (BX-Book-Ratings.csv)

---

🧹 Data Filtering Approach

To ensure high-quality recommendations, the dataset is filtered in two stages:

Step	Filter Condition	Reason	Before	After
1	Users with > 200 ratings	Keep only active readers with enough preference data	105,283	899
2	Books with ≥ 50 ratings	Keep only popular books for reliable similarity	160,269	742

Note: These thresholds remove sparse and noisy data, improving recommendation accuracy.

---

🏗 Architecture

Workflow:

1. Load & Clean Data – from CSVs (books, users, ratings)
2. Filter Users & Books – remove low-activity users & low-rated books
3. Create Pivot Table – books × users with ratings
4. Convert to Sparse Matrix – for efficiency
5. Train KNN Model – cosine similarity for nearest neighbors
6. Save Model & Data – serialized with pickle
7. Serve via Flask – API for /suggest & /recommend
8. Frontend UI – search & view recommendations

Tech Stack:

• Backend: Python, Flask
• ML: pandas, scikit-learn, SciPy
• Frontend: HTML, CSS, JavaScript
• Data: Book-Crossing dataset

---

⚙️ Installation & Setup

1️⃣ Clone the Repository

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

2️⃣ Install Dependencies

pip install -r requirements.txt

3️⃣ Train the Model

python train.py

4️⃣ Run the Flask App

python app.py