A hybrid book recommendation system that leverages deep learning, combining collaborative filtering and content-based approaches to deliver highly personalized and diverse book suggestions.
🔗 Try it Online (Hugging Face Space) 🔗 Open in Google Colab
Note: The interface and notebook content/documentation are in Portuguese (PT-BR).
This system harnesses the power of deep neural networks to analyze user behavior and book features, providing accurate, scalable, and diverse recommendations. Using PyTorch, it implements a neural collaborative filtering framework enhanced by content-based insights for optimal user experience.
GoodBooks-10k — one of the most comprehensive publicly available book rating datasets:
- Over 6 million ratings
- 10,000 distinct books
- 53,424 active users
- Ratings on a 1-5 star scale
- Programming Language: Python 3.12
- Deep Learning: PyTorch
- Data Processing: Pandas, Polars, NumPy
- Machine Learning: Scikit-learn
- Deployment: Gradio for interactive demo
- Cloud: Google Colab for easy experimentation
- Hybrid Recommendation: Fuses collaborative filtering and content-based techniques
- Deep Neural Network: Learns complex user-item interaction patterns
- Personalization: Delivers tailored suggestions for each user
- Scalability: Efficient handling of large-scale datasets and users
- Interactive Demo: Seamless web interface for exploration
- Open the Google Colab notebook
- Run all cells sequentially
- Experiment with parameters and observe recommendation changes
- Deploy the Gradio demo or integrate into your own system
- Embedding Layers: Learn latent factors for users and books
- Neural Collaborative Filtering: Multi-layer perceptron capturing nonlinear user-item relationships
- Matrix Factorization: Foundation for interaction modeling
- Hybrid Scoring: Combines collaborative and content-based scores for balanced recommendations
If you use this project in your research or work, please cite accordingly.