How to make inference after training a model?

Question

I have a question regarding the ALS model in ml. I'm building a movie recommendation using the MovieLens dataset(kaggle version).

After training the model, I got user matrix U and item matrix I. By doing U*I I can get the recommendations(by predicted ratings). I need to remove the training part of the dataset from the recommendations, as they have initially been rated by the users(or else as seen). But I'm having a hard time getting the predicted recommendations. I calculated the dot product, which is an array, but I'm having hard time trying to get the dataframe.

MY QUESTION is: How to get the recommendations dataset after the dot product to be able to outer join it with the training dataset to remove the already seen movies.

Any help would be appreciated