machine learning – Recommender system – Binary rating adding a condition

I want to implement a recommender system based on visualization. It will be collaborative filtering with a set of n users and each one will ha a set of m seen items.

In order to develop this, I was thinking on using the Jaccard similarity coefficient (if anyone has a better suggestion please let me know).

But, what if I want to add a condition to the result? For example, if the items are premium will appear first. This should be implemented at the end as a condition or there is another way?

Thanks in advance.