[💡SUG] Consider interactions frequency and not ratings

[gabarlacchi]

Hey there,
thanks for the amazing work here.

I would like to fully understand how to load a dataset (which also contains ratings value for each interaction) by considering the frequency of interaction user-item (same user,same item multiple times). In the sense that the model will look at how many times a user interacted with an item, no matters what the rating is every time.

Right now I load this kind of dataset also with rating column, but it's not clear to me where the model focus in order to make predictions or whatever linked to. If I load a dataset in which there are ratings, but also multiple interactions for the same user/items, what the model consider important?
Is this functionality already online, or there's a nice way to do it?

Best,
Gabriele

[TayTroye]

@gabarlacchi Hello!
Thanks for your attention. You may preprocess your dataset to incorporate interaction frequency between users and items and design corresponding models to effectively leverage this frequency information during training.