I try to implement content based filtering from scratch but i’m bit confuse on conputing w or b do i need to calculate w or b base on per users or per movies features just like collaborative filtering. or what ?

Hello @maDan_kD,

In a content-based filtering approach, you typically have two sets of weights: one set of weights for movies and another set of weights for users.

If you’re interested in diving deeper into these topics, I recommend enrolling in the “Deep Learning Specialization” course. It covers neural networks and related concepts in-depth and will provide you with a solid foundation for understanding and implementing recommendation systems.

Best regards,

Jamal

ok if we have users and users have 100 features so we need to calculate 100 weights and that 100 weights can be used for all users and same goes with movies?

Yes, in a content-based recommendation system with 100 features for both users and movies, you would calculate 100 weights for each movie and 100 weights for each user. These weights are specific to each user and movie to personalize the recommendations.

ok so if we have 110 users we need to calculate (100 weights per user because we have 100 features and if we have 110 users then (110 users)*(100 weights) is that correct?

Right, you would indeed need 100 weights per user, resulting in a total of 11,000 weights which is `(Number of Users * Number of Features = 110 users * 100 weights per user = 11,000 weights)`

to represent the preferences of all 110 users for all 100 features