Originally published at: https://developer.nvidia.com/blog/using-neural-networks-for-your-recommender-system/
This post is an introduction to deep learning-based recommender systems. It highlights the benefits of using neural networks and explains the different components. Neural network architectures are covered from the basic matrix factorization with deep learning up to session-based models.
Thank you for your help.
But I have some questions.
What features of users interactions dataset and products dataset should be used to make Embedding layers? Can you explain please?
It depends on your dataset. The categorical feature are often fed through an embedding layer. You can use all categorical item feature for item embeddings and all categorical user features for the user embedding - a good start is item id and user it. If you have multiple categorical features, you can concatenate them.