Testing out different Recommender algorithms on Amazon Reviews Dataset to find the best combination of accuracy and computation time.
Key Algorithms include:
- Factorization Machine using Tensorflow, using ratings,product attributes and text features extracted from the reviews
- Restricted Boltzmann Machine using Tensorflow
- Standard collaborative filterning algorithms. including KNN and Singular Value Decomposition
- Content-based recommender using KNN, with text features extracted from the reviews