https://github.com/alisatodorova/book-recommender

Comparison of Hybrid Book Recommender Systems: Matrix Factorization with Neural Networks vs. Neural Collaborative Filtering with Attention
https://github.com/alisatodorova/book-recommender

book-recommendation-system collaborative-filtering keras machine-learning matrix-factorization neural-network recommender-system tensorflow

Last synced: 27 days ago
JSON representation

Comparison of Hybrid Book Recommender Systems: Matrix Factorization with Neural Networks vs. Neural Collaborative Filtering with Attention

• Host: GitHub
• URL: https://github.com/alisatodorova/book-recommender
• Owner: alisatodorova
• Created: 2024-08-02T16:20:04.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2024-08-02T16:45:31.000Z (3 months ago)
• Last Synced: 2024-10-10T08:23:22.369Z (27 days ago)
• Topics: book-recommendation-system, collaborative-filtering, keras, machine-learning, matrix-factorization, neural-network, recommender-system, tensorflow
• Language: Jupyter Notebook
• Homepage:
• Size: 25 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Book-Recommender

**Title:** Comparison of Hybrid Book Recommender Systems: Matrix Factorization with Neural Networks vs. Neural Collaborative Filtering with Attention

**Abstract:** In this paper, we explore two advanced hybrid models for book recommendation systems: Matrix Factorization with Neural Networks (MF-NN) and Neural Collaborative Filtering with Attention (NCF-Attention). The goal of this research is to enhance prediction accuracy and generalization by leveraging deep learning techniques alongside traditional collaborative filtering methods. Our study employs the ”Book-Crossing: User review ratings” dataset from Kaggle to train, optimize, incrementally enhance, and then evaluate each of the models, focusing on their performance and scalability. Our findings indicate that while both models offer significant improvements over traditional approaches, the MF-NN model demonstrates superior performance in terms of accuracy and computational efficiency for our given dataset.

See **DL_Report.pdf** for the full report.

See folder **dataset** for the used dataset. File **test.ipynb** contains some visualisations and insights about the data.

Jupyter notebooks for each experiment can be seen in their corresponding folders, labeled **experiment#**, where # is 1 through 4.