Book Recommendation System

Overview

This project is a Book Recommendation System that provides users with book suggestions based on two approaches: a popularity-based (content-based) system and a collaborative filtering-based system. The goal is to recommend books that are relevant and engaging to users based on ratings data.

Application URL

https://book-recommendation-system-fd2b.onrender.com

Dataset

The dataset used for the purpose of this project can be accessed here.

The project utilizes three CSV files:

• books.csv: Contains information about the books.
• users.csv: Contains information about the users.
• ratings.csv: Contains user ratings for books.

Features

1. Popularity-Based Recommendation System

• Recommends the top 50 books with the highest average rating.
• Considers only books that have received at least 250 ratings to ensure reliability.

2. Collaborative Filtering-Based Recommendation System

• Recommends books based on user preferences and similarities.
• Filters:
  • Only considers users who have rated at least 200 books.
  • Only includes books with at least 50 ratings.
• Utilizes cosine similarity for building the collaborative filtering system.

3. Advanced Collaborative Filtering Using scikit-surprise

• Explored multiple algorithms for collaborative filtering:
  • SVD
  • KNNBasic
  • KNNWithZScore
  • NMF
• Finalized the use of SVD i.e Singular Value Decomposition to generate recommendations.
• Implementation Details for SVD:
  • Built a full training dataset using the scikit-surprise library.
  • Trained the SVD model on user-item interactions.
  • For a given book, calculated similarity scores between its latent factors and those of all other books using cosine similarity.
  • Ranked the books based on their similarity scores and retrieved the top 5 recommendations.
  • Each recommendation includes the book title, author, and cover image for enhanced user experience.

Installation

1. Clone this repository:

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

2. Navigate to the project directory:

   cd book-recommendation-system
   

3. Create a virtual environment (optional but recommended):

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

4. Install the required dependencies:

   pip install -r requirements.txt
   

Usage

1. Run the Streamlit application:

   streamlit run app.py
   

2. Open the provided local URL in your browser to use the system.

License

This project is licensed under the MIT License.