How does it work?

[liperrino]

Hi. I am a newbie in recommendation system, but I am really interested in what you have produced. Please, is it possible to have detailed explanations on how it works? I don't really understand your architecture. Thanks in advance.

[BeautifulBeer]

Hi, I cannot give you precise descriptions of this project at a fine-grained level, but I can give you some intuition of this project. As you know, two approaches are used in this project. One is Model-based CF(Collaborative Filtering) and the other is Content-based recommendation. Model-based CF can give a high-quality recommendation to users with the large-scale dataset, however, it suffers from the "Cold Start problem". The content-based method can handle this problem with a small dataset. Hence, we combine two approaches, model-based CF and content-based recommendation, to maintain the quality of recommendation whether the dataset is small or not. You can find more information with these keywords: "hybrid-recommendation", "model-based collaborative filtering", "content-based movie recommendation". I hope this comment will be helpful to you.

• dataset means the entire lists of movie ratings for users

[liperrino]

I understand. Please how do you obtain those latent files(latent_user.npy ...). I am building a recommendation system for books. That's why I want to know more. What happen in the mapper files?

[liperrino]

Thanks very much for your explanations.